Nuclear weapon

A nuclear weapon is an explosive device whose destructive force comes from reactions in the atomic nucleus rather than from chemical combustion. Fission weapons split heavy nuclei such as uranium-235 or plutonium-239 in a chain reaction, while thermonuclear weapons add a stage in which fusion of light isotopes releases far more energy. A single device can release the energy of thousands to millions of tonnes of TNT, which is why these weapons are treated as a separate category from conventional explosives. Their history is tied to the Second World War and the Manhattan Project, to the Cold War arms race, and to the diplomatic system that has tried to limit them since 1963.

The subject matters for physics, engineering, public health, law and international security. The entry explains the nuclear principles behind fission and fusion, the design and fuel choices that turned those principles into weapons, and the physical effects of a detonation. It then traces the development and spread of arsenals, the treaties that have constrained them, and the current state of global stockpiles, including the expiry of New START in February 2026. It also covers deterrence doctrine, command-and-control and accident risks, the medical and environmental consequences documented after tests and attacks, the legal and ethical arguments on both sides, and the modernization programmes and emerging technologies that shape the next phase of nuclear competition.

Nuclear physics behind the weapon

Binding energy and the release of energy

The energy in a nuclear weapon comes from how tightly the nucleons in a nucleus are held together. The binding energy per nucleon curve reaches a broad maximum near mass numbers of 56 to 62, where iron-56 and nickel-62 lie at roughly 8.79 MeV per nucleon. Beyond that peak the curve falls, because the growing number of protons produces ever-stronger Coulomb repulsion that the short-range strong force cannot fully offset. Nuclei on either side of the peak can therefore move toward it and release energy. Light nuclei do this by fusing, and heavy nuclei do it by splitting. The released energy equals the small loss of rest mass, as expressed by $E = \Delta m\,c^2$, the relation of mass-energy equivalence. Because the mass defect is only a fraction of a percent of the total mass, the energy per kilogram is enormous. Fission of a single uranium-235 nucleus releases about 200 MeV, which is millions of times more energy per kilogram of fuel than a chemical explosive releases.

Fission chains, neutron multiplication and critical mass

Uranium-235 and plutonium-239 fission most readily when they absorb a slow, or thermal, neutron. Each fission emits new neutrons. The average is about 2.4 for U-235 and about 2.9 for Pu-239. If, on average, at least one of these neutrons causes another fission, the process becomes a self-sustaining chain reaction. The effective multiplication factor, k, measures this balance. A mass in which k equals one is critical, and the smallest such mass for a given shape and density is the critical mass. Neutron reflectors and high density reduce the amount needed. A weapon must move from subcritical to strongly supercritical in a fraction of a microsecond. Only then can the chain reaction consume a significant share of the fuel before the core blows apart.

Prompt and delayed neutrons

Most fission neutrons appear almost instantly, but a small fraction is emitted later by fission products that undergo beta decay. For thermal fissions the delayed fraction is 0.65% for uranium-235 and only 0.21% for plutonium-239. In a power reactor this delay is what makes control possible. The relatively slow timescale of delayed neutrons allows reactor power to be controlled through the gradual movement of control rods. Reactors are therefore designed to operate in a delayed-critical mode and are provided with safety systems to prevent them from ever achieving prompt criticality.

A weapon works in the opposite regime. The aim is to reach prompt criticality, where the chain reaction is driven by prompt neutrons alone, and to complete the energy release within about a microsecond. The smaller delayed fraction of plutonium matters here: in plutonium systems prompt criticality is reached after a far smaller reactivity increase than in uranium systems. Designers must therefore manage the risk that a plutonium core becomes supercritical too early and fizzles, a concern that shapes assembly speed and initiator timing.

Deuterium–tritium fusion

Fusion requires nuclei to overcome their mutual electrostatic repulsion, which demands extreme conditions. The deuterium–tritium reaction is the easiest to ignite and is the one used in weapons. One deuterium nucleus fuses with one tritium nucleus, giving one helium-4 nucleus, one free neutron and 17.6 MeV of total energy. Its ignition temperature is about 160 million kelvin, compared with about 15 million kelvin for the proton-proton chain in the core of the Sun. Temperature alone is not enough. Fusion also requires sufficient pressure or density and a confinement time long enough for the reaction to proceed. The fuel must be in a plasma state, with electrons stripped from the nuclei. In a thermonuclear weapon, these conditions are supplied by a fission primary, which is why the physics of fission and fusion is inseparable in weapon design.

Fissile and fusion materials and the fuel cycle

Natural uranium is about 0.7 percent uranium-235, the isotope that sustains a fission chain reaction with slow neutrons. The rest is mostly uranium-238, which captures neutrons without readily fissioning. Producing weapon material therefore requires raising the U-235 fraction through isotope separation. Because the two isotopes differ by only three neutrons in mass, separation relies on small physical differences. The first industrial method, gaseous diffusion, forced uranium hexafluoride gas through porous barriers, so that lighter molecules containing U-235 passed through slightly more readily. The Oak Ridge K-25 plant, built during the Manhattan Project, used this approach and consumed very large amounts of electricity. Gas centrifuges, which spin the same gas so that heavier molecules concentrate near the rotor wall, later replaced diffusion because they need far less energy per unit of separative work and occupy much less space.

Enrichment levels define the main categories of uranium. Low-enriched uranium (LEU) is generally below 20 percent U-235 and is typical of power-reactor fuel, which usually contains a few percent. Highly enriched uranium (HEU) is 20 percent or more, and weapon-grade material is typically around 90 percent U-235. The step from reactor fuel to weapon material is the most demanding part of the process, and it is the reason enrichment facilities receive the closest attention in nonproliferation policy. A gun-type device needs HEU, whereas implosion designs can tolerate more varied material, a point developed in the design sections.

Plutonium does not occur in useful quantities in nature. It is made in reactors when uranium-238 captures a neutron and becomes uranium-239, which undergoes two beta decays through neptunium-239 to plutonium-239. Fuel removed after a short irradiation contains mostly Pu-239. The longer fuel remains in the core, the more Pu-239 captures a further neutron and becomes Pu-240. Pu-240 has a high spontaneous fission rate, so it releases neutrons that can start a chain reaction before the core is fully assembled, causing a weak "fizzle." This is why reactor-grade plutonium, with a high Pu-240 share, is poorly suited to weapons. Weapon-grade plutonium is commonly defined as containing below about 7 percent Pu-240, a threshold produced by short irradiation cycles in dedicated production reactors.

Fusion weapons burn deuterium and tritium, a reaction that yields a helium-4 nucleus and a neutron and releases about 17.6 MeV per event. Deuterium is abundant, but tritium is not. Its half-life is about 12.3 years, so a stockpile loses roughly 5.5 percent of its tritium each year as it decays to helium-3. Tritium must therefore be produced continuously. The main route is neutron irradiation of lithium-6 in reactors, in which the reaction Li-6 + n gives He-4 + T. Because natural lithium is mostly lithium-7, the lithium must first be enriched in lithium-6. The United States used dedicated reactors at Savannah River for this purpose and later relied on commercial reactors such as TVA's Watts Bar. Thermonuclear secondaries also contain lithium-6 deuteride, which generates tritium in situ from neutrons produced during the burn, so that the device does not need to store large amounts of tritium gas.

Material accounting is inherently difficult. The core of an implosion warhead, the pit, contains a few kilograms of plutonium, while secondaries contain uranium components, including some HEU in certain designs. These parts are small, dense, classified and can be stored in disassembled form, which makes them hard to count from outside. Civil material is tracked under the nuclear fuel cycle safeguards of the International Atomic Energy Agency, but military stockpiles fall outside those obligations unless states volunteer them. A proposed Fissile Material Cut-off Treaty, which would ban new production of fissile material for weapons, has been debated for decades without being concluded. Without such a treaty, verifying how much weapon-grade material exists, and how much has been retired, depends largely on national declarations and indirect estimates.

Basic weapon designs: gun-type and implosion

Fission weapons rely on one principle: a mass of fissile material must be assembled quickly into a supercritical configuration so that a chain reaction of nuclear-fission proceeds before the material blows itself apart. Early designers found two main ways to achieve this. Which one was feasible depended mainly on the isotopes available, so the choice between them was less a matter of preference than of physics.

Gun-type assembly

The gun-type principle was used in the uranium bomb "Little Boy," dropped on Hiroshima on 6 August 1945. A propellant charge fired one subcritical piece of uranium down a gun barrel into a second subcritical piece, forming a supercritical mass. The projectile was a hollow cylinder and the target a solid one, and the two pieces were held apart until the moment of firing. Little Boy contained roughly 64 kilograms of highly enriched uranium, and its yield is usually estimated at 13 to 16 kilotons. The design was highly inefficient, because only a small fraction of the fuel fissioned before the device disassembled. Its designers judged it reliable enough that the bomb was never tested beforehand. Production of purified uranium-235 had consumed nearly all of the material available at the time, so no comparable device could be built for a test.

The gun method only works with highly enriched uranium. Uranium-235 has a low spontaneous fission rate, so the pieces can be brought together in the short window before stray neutrons start a chain reaction. Reactor-bred plutonium, by contrast, contains plutonium-240, which fissions spontaneously at a much higher rate. This would cause premature detonation, or "fizzling," before the assembly is complete. For this reason the plutonium gun design known as Thin Man was abandoned in 1944 after technical difficulties emerged. Gun-type weapons were later largely superseded, and the designs built in the United States were dismantled.

Implosion

The implosion method was used for the plutonium device tested at Trinity in July 1945 and for "Fat Man," dropped on Nagasaki on 9 August 1945. A subcritical sphere of plutonium is surrounded by a dense tamper and an outer shell of chemical high explosives. The explosives are shaped into "explosive lenses," arrangements of fast and slow explosives that focus the detonation wave inward into a nearly spherical shock. All the lenses must detonate at the same instant, which requires a synchronized firing system. The inward shock compresses the core to a higher density, at which it becomes supercritical. Fat Man was about 1.5 metres in diameter, far bulkier than the gun design, but the method used far less fissile material for a comparable yield.

Implosion became the standard for fission weapons for several reasons. It tolerates the higher spontaneous fission rate of reactor-grade plutonium. It makes more efficient use of scarce fissile material. It can be scaled to smaller packages, which made it suitable for missiles, artillery shells and bombs of varying size. Variants with different geometries were developed for devices that had to fit inside small diameters.

Neutron initiators and tampers

The tamper is a dense layer, often of natural or depleted uranium, that surrounds the core. Its inertia holds the core together long enough for multiple fission generations to occur, and it reflects escaping neutrons back into the fuel. Without it, a larger share of the material would be wasted. The tamper can reduce the critical mass of uranium-235 to around 20 to 25 kilograms. Other reflector materials, such as beryllium, serve a similar role.

A neutron initiator supplies the first burst of neutrons needed to start the chain reaction. An early design consisted of a pellet of polonium-210 and beryllium, which releases neutrons when crushed. The initiator is placed at the centre of the core, so that the burst arrives when the core is at maximum compression rather than earlier, when it could cause a fizzle. Timing of the burst is therefore as important as its strength.

Boosted fission

Boosting adds a small quantity of fusion fuel to the hollow core of a fission device. Edward Teller had conceived the concept by mid-1943. A mixture of deuterium and tritium gas, or tritium alone, is stored and released into the core. The fusion reactions emit a burst of high-energy neutrons that induce fission far faster than the normal chain reaction can. This raises the yield from a given amount of fissile material, or achieves the same yield with less of it. The fusion step itself contributes only a small share of the energy, perhaps about 1 percent. Its main value is miniaturisation, since a boosted primary can be fitted into a compact warhead, where it can ignite a thermonuclear secondary. Boosting was developed at Los Alamos between late 1947 and late 1949, and it is credited with much of the increase in fission-weapon efficiency over the following decades. Because tritium decays with a half-life of about twelve years, boosted devices need periodic replacement of their tritium supply.

Thermonuclear weapons and staging

Thermonuclear weapons add a fusion stage to a fission trigger, and nearly all modern designs follow the Teller–Ulam configuration. A fission "primary" fires first. Its X-rays are guided through a radiation case to a separate "secondary" that holds fusion fuel, and that energy compresses the secondary before fusion begins. The fuel is primarily lithium-6 deuteride, and during detonation, neutrons convert lithium-6 to helium-4 plus tritium, so the reaction chain itself produces tritium for fusion. The design is a sequence of physical steps. Inside the case, the space is filled with a light material that becomes a hot plasma, which re-radiates X-rays onto the secondary's outer shell. That shell ablates, and the recoil drives the fuel inward to very high density. A fission "spark plug" embedded in the fuel then heats it further. The outer layers of the secondary are opaque to X-rays, typically made of lead or uranium-238. If the outer layer is uranium, it can capture neutrons produced by the fusion reaction and undergo fission itself, increasing the overall explosive yield.

The concept took shape quickly. Edward Teller and Stanisław Ulam devised the concept in January 1951 for the United States nuclear weapons program, although the precise division of credit remains disputed. Ulam's idea of a bomb within a container was refined by Teller, who proposed using the primary's radiation rather than its neutron stream to compress the secondary. The US Greenhouse George test in May 1951, the world's first artificial thermonuclear fusion, validated the radiation implosion principle.

The first full test came on 1 November 1952. The first Ivy shot, codenamed Mike, was the first successful full-scale test of a multi-megaton thermonuclear weapon using the Teller-Ulam design. It yielded 10.4 megatons, almost 500 times the yield of the bomb dropped on Nagasaki, and eight megatons of the yield were from fast fission of the uranium tamper. The device used liquid deuterium and weighed some 82 tons, so it could not be delivered. The next US step was Castle Bravo, a dry, solid-fuel device on 1 March 1954. It yielded 15 megatons of TNT, 2.5 times the predicted 6 Mt, due to unforeseen additional reactions involving lithium-7.

The Soviet path was less direct. The 1953 test was based on the layer-cake design, which is not true radiation-imploded staging. In late 1953, Viktor Davidenko achieved the first breakthrough, that of keeping the primary and secondary parts of the bombs in separate pieces ("staging"). Sakharov and Yakov Zeldovich then developed the use of X-rays from the fission bomb to compress the secondary before fusion, in the spring of 1954. That design was tested as RDS-37 in November 1955, which was the world's first air-dropped multi-stage thermonuclear test. The 1952–1953 tests therefore demonstrated the principles, but the mature staged design was confirmed only in 1955.

Yield can be increased by enlarging the secondary and adding stages. The Soviet Tsar Bomba, detonated on 30 October 1961, was a three-stage device. Its design could reach up to 100 megatons when fully loaded with nuclear fuel. For the test, designers reduced the yield by 50% by eliminating the fast fissioning of the uranium tamper by the fusion neutrons, and 97% of the energy came from fusion reactions. Published figures differ: bhangmeter results and other data suggested the bomb yielded around 58 Mt, but Soviet scientists later revealed that their instruments indicated a yield of 50 Mt.

Multi-megaton devices are rarely built for field use for several reasons. Weight and size limit delivery, which is why the early Mike device was impractical. The American B41 offered a useful comparison: it had a mass of 4800 kg and a yield of 25 Mt, which equates to the highest yield-to-weight weapon ever mass-produced. Even so, both the United States and the Soviet Union stopped prioritizing blast yield at about this point, instead prioritizing missiles that could deploy multiple nuclear weapons at once. Accuracy and target coverage generally favor several smaller warheads over one very large one, and very high yields produce more fallout and collateral damage.

Classes of weapons and yield ranges

Strategic and non-strategic weapons

Nuclear weapons are usually classified by mission and range, not only by design. Strategic weapons are intended for targets far from the battlefield, such as cities, leadership sites, industrial centres and missile fields, and are normally delivered by intercontinental ballistic missiles, submarine-launched ballistic missiles or long-range bombers. Non-strategic weapons, often called tactical or theatre weapons, are meant for use against military forces in a regional conflict. They are typically carried by artillery, short-range missiles, and fighter or bomber aircraft, and they generally have shorter ranges and lower yields. The boundary is not sharp. A warhead's role depends on its range, its planned target and its operational doctrine, and some designs can serve either role. The distinction matters for arms control because treaties such as New START have mainly limited deployed strategic warheads, leaving most non-strategic arsenals outside numerical caps, which makes them harder to count and verify.

Enhanced radiation weapons

Enhanced radiation weapons, popularly called "neutron bombs", are designed to release a larger share of their energy as fast neutrons and less as blast and heat. The concept is usually credited to Samuel T. Cohen, who worked on it at the Lawrence Livermore National Laboratory in 1958, and the idea was tested in 1963 at a Nevada underground site. The rationale was to attack armoured vehicles and troops while limiting physical damage to nearby buildings and populations, particularly in Western Europe where NATO planners expected any battlefield nuclear use to occur. Two American systems carried this feature. The W70-3 warhead, for the short-range Lance missile, had a yield of roughly 1 kiloton and was produced in 1981–83, with 380 built. The W79 was an artillery shell for the eight-inch gun, with an enhanced-radiation mode that could be switched on or off. Deployment was politically contentious. President Jimmy Carter postponed the programme in 1978 after protests, Ronald Reagan restarted production in 1981, and a 1985 congressional order required future W79s to be built without the enhanced-radiation feature, with existing units modified. The W70-3 was retired by 1992 and the last W79 was dismantled by 2003.

Critics argued that the lower yield and battlefield utility of such weapons would make nuclear use seem more acceptable and so lower the threshold for nuclear war. Supporters argued that the weapons offered a more discriminating option than older battlefield warheads. Analysts also note that yield figures can be misleading for these designs, because much of their harm comes from prompt radiation rather than blast.

Low-yield weapons and the deterrence debate

The modern debate over small warheads centres on the American W76-2, a low-yield variant of the W76-1 warhead for the Trident II (D5) submarine-launched missile. It was first proposed in the Trump administration's 2018 Nuclear Posture Review. Its yield is classified, but analysts such as the Federation of American Scientists estimate roughly 5 to 7 kilotons, compared with about 90 kilotons for the W76-1, achieved mainly by removing or disabling the secondary thermonuclear stage. The Navy deployed it on the USS Tennessee in late 2019, and the Pentagon confirmed the deployment in February 2020. Estimates of production are around 50 warheads, converted from existing stock.

The review justified the weapon as a response to Russia's large tactical arsenal and to a perceived doctrine of "escalate to de-escalate", arguing that the United States needed a credible option short of a full strategic exchange. Critics, including several members of the US Congress, countered that low-yield weapons could blur the line between conventional and nuclear conflict and make nuclear use more thinkable. They also pointed out that an adversary observing a Trident launch could not tell whether the warhead was low or high yield, which could raise the risk of miscalculation, and that existing bomber-delivered and air-launched cruise missile options already gave similar flexibility. The debate thus turns on a basic question in deterrence theory: whether limited options strengthen deterrence by increasing credibility, or weaken it by lowering the cost of initiating nuclear use.

Radiological dispersal devices are not nuclear explosives

A radiological dispersal device (RDD), commonly called a "dirty bomb", combines a conventional explosive with radioactive material. Its explosive power comes entirely from the chemical explosive, and the radioactive material is scattered around the blast site. Because there is no fission or fusion reaction, the device does not release nuclear energy and cannot produce the blast, heat or long-range fallout of a nuclear weapon. Regulators such as the US Nuclear Regulatory Commission describe the radiation from such a device as typically spread over a few blocks or miles, whereas the fallout from a nuclear explosion can cover hundreds or thousands of square miles. Most casualties from an RDD would be caused by the conventional blast, and the radiological exposure would rarely be lethal. Its main effects would be contamination, the cost and time of decontamination, and fear. For these reasons it is often described as a weapon of mass disruption rather than mass destruction. No such device has been detonated in an attack, although the sources describe several tests and many security analyses.

Physical effects of detonation

Blast

A nuclear explosion releases energy so fast that the surrounding air is pushed outward as a shock wave. The key measure is overpressure, the amount by which the pressure at the shock front exceeds ambient atmospheric pressure. Overpressure of a few pounds per square inch is sufficient to destroy typical wooden houses. Blast damage is often described in terms of a radius of destruction, often taken as the zone of 5-pound-per-square-inch overpressure.

Blast reach scales with yield according to the Hopkinson–Cranz, or cube-root, law. The distance at which a specific overpressure is produced by air blast scales as the cube root of the yield. In practice, a given overpressure reaches about 0.7 km from a 1-kiloton air burst, 3.2 km from 100 kilotons and 15 km from 10 megatons. Because of this scaling, one large weapon covers far less area than several smaller ones with the same total energy, which is one reason for multiple-warhead designs (see nuclear-weapon delivery sections).

Burst height changes the pattern. Lowering the height of an air burst for a given yield brings the strongest blast closer to ground zero, but the total area of coverage for blast effects is also markedly reduced. The overpressure phase also lengthens with energy: for a 1-kiloton air burst it lasts a fraction of a second, while for a 10-megaton burst it can persist for several seconds.

Thermal radiation and fireballs

The energy release heats the device and the surrounding air to extreme temperatures. Temperatures of several tens of millions of degrees centigrade develop in the immediate area of the detonation, in contrast to the few thousand degrees of a conventional explosion. The heated, ionized material forms a fireball. Within a millisecond after detonation, the diameter of the fireball from a 1 megaton air burst is 150 m, and it grows to a maximum of 2200 m within 10 seconds.

Roughly a third or more of the energy emerges as thermal radiation. Thermal radiation accounts for between 35 and 45 percent of the energy released, depending on the yield. The radiation causes burns and eye injuries, and the risk extends beyond the blast zone in clear weather. Thermal pulses can also ignite material at great distances: thermal radiation can ignite wood-frame buildings, vegetation and other combustible materials at significant distances from the site of the explosion. In large urban areas, many separate fires can merge into a firestorm, which increases the area affected.

Prompt radiation and fallout

Prompt ionizing radiation, mainly gamma rays and neutrons, is released within roughly the first minute. Its intensity falls off steeply with distance, and it is largely a hazard within the immediate blast zone. Delayed radiation comes from radioactive material created after the explosion. Fission products, along with neutron-activated elements in soil and structures, form the basis of fallout. Fallout is produced by fission products and neutron-induced radionuclides in surrounding materials such as water, soil, structures and nuclear device debris.

The amount and distribution of fallout depend strongly on how the weapon is detonated. Ground bursts or explosions closer to the ground disturb large amounts of soil and draw it into the fireball. The radioactive isotopes condense onto this debris, which is heavier than air and settles back to the ground within minutes or days. A surface burst thus produces local fallout that can be a hazard over a much larger downwind area than the blast and thermal effects cover. Fallout from an airburst is generally smaller, though even apparent airbursts can still cause significant fallout, due to neutron activation. Rain can wash the particles out of the cloud, which shortens the distance they travel but concentrates the contamination in patches.

Fallout is conventionally divided into local fallout, a plume of concentrated material near the detonation within minutes to hours, and global fallout, a diffuse, low-level deposition over large areas. The Soviet Tsar Bomba test of 1961 illustrates how height matters: its fireball was buoyantly boosted upwards by its reflected shockwave, which prevented much of the mixing that would have produced local fallout.

Electromagnetic pulse from high-altitude bursts

A detonation well above the atmosphere produces little blast or thermal effect at the ground, but its gamma rays interact with the upper atmosphere and generate a rapid, intense pulse of electromagnetic energy over a large region. The mechanism involves fast electrons released by gamma-ray interactions being deflected by Earth's magnetic field. The resulting pulse can overload unshielded electronics, communications equipment and sections of the power grid across a wide area, although the precise magnitude depends on altitude, yield and the shielding of target systems.

Hypothesized nuclear winter

Large fires could loft soot into the upper atmosphere, where it might block sunlight for an extended period and cool the surface. This idea, advanced in a 1983 study known as TTAPS, is called nuclear winter. Early models suggested severe and prolonged global cooling, but later simulations produced a wide range of results, and the scale of any effect remains contested. The hypothesis is therefore best understood as a plausible but uncertain consequence of large-scale nuclear war, not a settled prediction.

Delivery systems and the nuclear triad

A nuclear warhead is only useful if it can reach its target, and the choice of delivery system determines how fast, how accurate and how survivable a force is. Strategic nuclear forces of the major powers are organized around a nuclear triad, which combines land-based missiles, submarine-launched missiles and bombers. In the U.S. case, the 2020 revised Nuclear Matters Handbook listed 14 ballistic missile submarines carrying 240 submarine-launched missiles, 400 land-based ICBMs, and 60 nuclear-capable heavy bombers. Each leg has different strengths, and the main logic of the triad is that an adversary cannot destroy all three in a single first strike (each leg provides distinct advantages, and together they form a deterrent that cannot feasibly be eliminated in one blow).

Intercontinental ballistic missiles

An intercontinental ballistic missile (ICBM) is a long-range, land-based missile. Most are housed in hardened silos. Because they are dispersed across territory and placed in blast-resistant silos, they are well protected against a first strike, although critics argue their fixed position makes them more vulnerable. Other states, notably Russia, have also used road-mobile launchers, which make a missile harder to locate but require more operational effort. ICBMs respond very quickly: they can be launched within minutes on presidential authority. Their drawback is their static nature. They offer few options for signalling intentions during an individual crisis, since they cannot be moved and no spare missiles or silos exist.

Multiple warheads on a single missile

Many ICBMs and submarine-launched missiles carry several warheads through a multiple independently targetable reentry vehicle (MIRV) payload. Unlike a traditional missile that carries one warhead, a MIRV-equipped missile can carry several. After the booster burns out, a post-boost vehicle, often called the bus, carries the re-entry vehicles, which are then released separately, and warheads can be released at different speeds and in different directions. The concept is distinct from the older multiple reentry vehicle, in which warheads are dispersed without individual aiming (the MIRV differs from the MRV, which deploys multiple warheads against a single target). MIRVs increase the number of targets a single launcher can threaten, and they complicate defences. Under the treaty limits in force for the United States, the Minuteman III is capable of carrying three warheads but is currently limited to one per treaty.

Submarine-launched ballistic missiles

Ballistic missile submarines (SSBNs) are generally regarded as the most survivable leg. A portion of the fleet is always at sea, which makes these submarines very hard to track. That survivability underpins the second-strike capability on which deterrence rests. The U.S. sea leg relies on Ohio-class submarines carrying Trident II D5 missiles, and the Columbia class is being built to replace the existing fleet beginning in 2031. Analysts often describe the sea leg as the core of the triad, since it is the least vulnerable to a surprise attack.

Strategic bombers and air-launched cruise missiles

Bombers form the third leg. Their main advantage is flexibility: they can change their armaments and deployments and also carry out conventional missions. A bomber can be recalled or redirected after launch, which ICBMs and submarine missiles cannot. The U.S. heavy bomber fleet in the early 2020s included 20 B-2 bombers and 40 nuclear-capable B-52 bombers, with the B-21 under development. Bombers can carry gravity bombs or stand-off air-launched cruise missiles, which allow a crew to strike from outside an air defence zone. Russia's Tu-22M3 bomber, for example, can deliver the Kh-22 air-launched cruise missile (all of these dual-capable aircraft can deliver nuclear gravity bombs, and the Tu-22M3 can also deliver air-launched cruise missiles).

Tactical delivery by aircraft, artillery and missiles

Non-strategic systems are designed for battlefield use and are generally lower in yield. Their delivery means include short-range ballistic missiles, artillery shells, air-dropped bombs, cruise missiles, torpedoes, and even landmines. Nuclear artillery is a subset of limited-yield tactical weapons launched from the ground at targets on the battlefield, and tactical ballistic missiles fill the gap between conventional rocket artillery and longer-range missiles, carrying heavy payloads deep behind enemy lines. Aircraft delivery poses its own problems. Early tactical bombs required special fuzing that would permit fighter pilots to deliver nuclear bombs at low altitudes and still escape the blast. Modern delivery must also contend with air defences, which is why many systems use standoff range, low-altitude flight or manoeuvring reentry. The U.S. B61-12 bomb is an example of a modern tactical weapon whose output can be set to anywhere from 0.3 to 50 kilotons depending on the target. The boundary between tactical and strategic is not purely technical: tactical weapons use the same underlying physics as their strategic counterparts but are engineered for lower yields and smaller packages.

Origins and the Manhattan Project

The American nuclear programme began as a response to fears that Nazi Germany might build such a weapon first. Hungarian refugees Leo Szilard, Eugene Wigner, and Edward Teller persuaded Einstein to warn President Franklin D. Roosevelt about this possibility and to urge a comparable effort in the United States. Szilárd wrote the text. It was most likely written by Leo Szilard, the scientist who invented the chain reaction, while Einstein's signature gave it weight. The letter is dated 2 August 1939. It pointed to work by Enrico Fermi and Szilárd, and noted that German authorities had already halted uranium sales from the Czech mines they had seized, Germany has actually stopped the sale of uranium from the Czechoslovakian mines which she has taken over. It warned that uranium could become the basis for "extremely powerful bombs", and it asked the government to fund chain-reaction research and keep in close contact with the physicists doing it. Einstein later regretted his role. He told an interviewer that he would not have acted had he known the Germans would fail, a remark that reflects how the German programme was judged after the war.

The letter did not produce immediate action. Alexander Sachs, who delivered it, met with Roosevelt twice - first on October 11, 1939, six weeks after the war had begun in Europe. At that meeting the president was not persuaded that the U.S. government should get involved. Germany's invasion of Poland on 1 September 1939 had already begun the European war, and the United States remained neutral. Roosevelt's response was gradual. He set up the Advisory Committee on Uranium, marking the first formal step by the U.S. government toward organized nuclear research. Support grew slowly through 1940 and 1941, and the effort accelerated sharply after the United States entered the Second World War in December 1941.

From 1942 the programme was run as the Manhattan Project. From 1942 to 1946, the project was directed by Major General Leslie Groves of the U.S. Army Corps of Engineers. Scientific design work centred on Los Alamos, where nuclear physicist J. Robert Oppenheimer was the director of the Los Alamos Laboratory that designed the bombs. Production was spread across three secret sites. Los Alamos, New Mexico designed the weapon. Oak Ridge, Tennessee enriched uranium. Hanford, Washington produced plutonium. The industrial scale was very large. The Manhattan Project employed nearly 130,000 people at its peak and cost nearly US$2 billion, and over 90% of the cost was for building factories and producing the fissionable materials, with less than 10% for development and production of the weapons. Oak Ridge alone grew dramatically, with its population rising from approximately 3,000 in 1941 to as many as 75,000 people at the height of the project. The chemical company DuPont was persuaded to construct and manage the project's plutonium production facilities. Hanford faced severe staffing problems, with a severe labor shortage (up to 50 percent) existed for the first 14 months of construction.

The Trinity test was the decisive technical milestone. On 16 July 1945, using a steel tower, scientists hoisted and suspended Gadget 100 feet into the air, and at 5:29 a.m. on July 16, 1945, the Trinity Test began. Oppenheimer had expected a yield of about 0.3 kilotons, but the resulting blast equated to roughly 21 kilotons of TNT. The test was considered essential because the implosion technology device was so complex. It tested the plutonium implosion design, the same technology that would be deployed in the weapon that exploded over Nagasaki. The uranium gun-type device intended for Hiroshima had never been tested in full. There was only enough weapons-grade uranium available for one bomb, and confidence in the gun-type design was high, so on July 14, 1945, most of the uranium bomb ("Little Boy") began its trip westward to the Pacific without its design having ever been fully tested. Residents near the test site were not warned: nearly half-a-million people were living within a 150-mile radius of the explosion, with some as close as 12 miles away. None were warned or evacuated by the US government ahead of time.

The bombings followed quickly. On 6 August, 1945, the United States dropped a nuclear weapon on the city of Hiroshima, Japan and three days later on 9 August, the US dropped another nuclear bomb over the city of Nagasaki. Estimates for Hiroshima killed an estimated 140,000 people in all, most of whom were civilians. At Nagasaki, some 40,000 people, again mostly civilians, died instantly. President Harry Truman argued that the bombings were necessary to compel Japan's surrender and to avoid what likely would have been a far deadlier U.S. invasion. Roosevelt had died earlier, and he left behind no clear plans for how the United States should use an atomic bomb.

The decision to use the weapons was contested at the time and remains so. Among the scientists, Szilárd morally objected to the use of an atomic bomb in warfare and drafted a petition for President Truman demanding the Japanese be warned before the atomic bomb was used. Szilard's letter was never received by the President and the 12 signatories were criticized by Manhattan Project leadership. Historians and ethicists still disagree about whether the bombings were necessary, and Truman's decisions, and particularly the decision to bomb Nagasaki, remain hotly debated. The original rationale for the project also became moot, since Germany never developed an atomic weapon.

Cold War arms race and spread to more states

The Soviet Union's first nuclear test, RDS-1, was fired at the Semipalatinsk test site in Kazakhstan on 29 August 1949. It was a plutonium implosion device with an estimated yield of about 22 kilotons, closely modelled on the American "Fat Man" bomb. Western governments were caught off guard, and the test ended the American monopoly and started the arms race that defined the Cold War. The Soviet design team soon produced a second device using highly enriched uranium, RDS-2, which was first tested in 1951. The first Soviet thermonuclear test followed in November 1955, with the thermonuclear weapon programme already under way in both countries.

Expansion of the American and Soviet arsenals

Through the 1950s and 1960s both superpowers built up their stockpiles rapidly. The American arsenal reached its largest size in the mid-1960s, at more than 31,000 warheads, and then fell back to the low 20,000s during the 1980s. The Soviet build-up was more gradual but did not stop: the Soviet inventory overtook the American one in 1978 and peaked at more than 40,000 warheads in 1986. Global totals peaked the same year, at roughly 70,300 warheads according to Federation of American Scientists (FAS) estimates. Some other tallies put the figure for the operational military stockpile alone at about 64,000, because the 70,300 figure includes retired warheads still awaiting dismantlement. The growth of these arsenals was driven by the search for a secure second-strike capability, which required more warheads, more delivery systems, and more kinds of weapons.

Entry of further states

Britain, France and China joined the nuclear club in the 1950s and 1960s. Britain conducted its first test, Operation Hurricane, at the Monte Bello Islands off Australia on 3 October 1952. France tested its first device, Gerboise Bleue, at Reggane in French Algeria on 13 February 1960. China's first test took place at Lop Nur in Xinjiang on 16 October 1964. The programmes that followed took different routes. India, which had never joined the Non-Proliferation Treaty, detonated a device in May 1974, codenamed "Smiling Buddha", at Pokhran. That test prompted Pakistan to accelerate its own secret programme, and the two countries carried out a round of tests in May 1998. Israel is widely assumed to possess nuclear weapons but has never tested them publicly and maintains a policy of nuclear ambiguity. North Korea announced its withdrawal from the NPT in 2003 and conducted its first nuclear test in October 2006 at Kilju. Israel, India and Pakistan never joined the NPT, while North Korea left it after having been a member.

The Cuban Missile Crisis

The most dangerous moment of superpower confrontation came in October 1962. American reconnaissance revealed Soviet medium-range missiles being installed in Cuba, and the two sides were brought close to nuclear war during a thirteen-day standoff. Washington responded with a naval quarantine around the island, while Moscow refused to withdraw the missiles at first. The crisis was resolved when Nikita Khrushchev agreed to remove the Soviet missiles in exchange for a United States pledge not to invade Cuba, together with a secret American commitment to withdraw its Jupiter missiles from Turkey. The episode accelerated efforts to introduce crisis communication channels and later arms control.

Peak and decline

Since the 1986 peak, global inventories have fallen by more than 80 per cent, but most of this reduction took place in the 1990s, and the pace has since slowed. FAS estimated the total at about 12,321 warheads at the beginning of 2026. Much of the continuing decline reflects the dismantlement of retired warheads, while the number of warheads in military stockpiles is rising again. Hence the post-Cold War reduction was not a steady trend, and the smaller arsenal of today is no longer decreasing in a straightforward way.

Nuclear testing and its legacy

Atmospheric tests and global fallout

Between 1946 and 1958, the United States set off 67 devices at the Pacific Proving Grounds, the U.S. military conducted 67 nuclear tests in the Pacific Proving Grounds from 1946 to 1958. The most consequential of these was the Castle Bravo shot at Bikini Atoll on 1 March 1954. The device was meant to test lithium deuteride as a fusion fuel, but a miscalculation resulted in the explosion being over twice as large as predicted, with a total explosive force of 15 megatons of TNT. Roughly 10 megatons of that came from fission of the natural uranium tamper, but those fission reactions were quite dirty, producing a large amount of fallout.

Shifting winds carried the debris far beyond the test area. Combined with the much larger than expected yield and an unanticipated wind shift, radioactive fallout spread into unexpected areas. Inhabited atolls to the east were hit, and the military did not evacuate residents until two days after the test occurred. Children on Rongelap played in the ash, which residents took for snow. A Japanese tuna boat, the Daigo Fukuryu Maru, was also contaminated, and many of the crew became ill, with one fatality. The incident created outrage around the world and pressure to ban nuclear weapons tests, and the U.S. agreed to pay more than $15 million in compensation to the Fukuryu Maru survivors.

The Soviet Union's Tsar Bomba was a different kind of event. Moscow ordered the project in July 1961 as part of the Soviet resumption of nuclear testing after the Test Ban Moratorium. The device was detonated in the atmosphere at 11:32 Moscow Time over the Mityushikha Bay Nuclear Testing Range in the northern Arctic Circle. Its yield was long reported as about 58 megatons, which was the accepted yield in technical literature until 1991, when Soviet scientists revealed that their instruments indicated a yield of 50 Mt. The design could have gone higher. It could theoretically have yielded as much as 100 megatons, but it would have resulted in a dangerous level of nuclear fallout, so the uranium tamper was replaced with lead for the test, which roughly halved the output.

Underground testing after 1963

The 1963 Partial Test Ban Treaty prohibited explosions in the atmosphere, underwater and in outer space, but not underground. Testing therefore moved below ground, where debris was largely contained, though venting and leakage still occurred. The main sites were the Nevada Test Site in the United States, the Semipalatinsk range in Soviet Kazakhstan, the Arctic archipelago of Novaya Zemlya, and the Pacific islands. Novaya Zemlya, an Arctic archipelago in the Barents Sea, served as the Soviet Union's northern nuclear testing site from 1955 to 1990. The site hosted 224 nuclear tests. France later used Mururoa and Fangataufa in French Polynesia, and the Chinese programme used Lop Nur. Underground work was less visible than the atmospheric tests but still produced local contamination and displacement, especially where test shafts were drilled near populated areas or fragile geology.

Detection and verification

Detection of tests predates the treaty era. Starting in the late 1940s, the United States developed a capability to monitor atmospheric nuclear tests and was successful in detecting the first, unsuspected Soviet nuclear test in late August 1949 by routine air sampling over the Pacific Ocean. Today the verification architecture of the Comprehensive Nuclear-Test-Ban Treaty Organization rests on the International Monitoring System, which comprises four technologies: seismological, radionuclide, hydroacoustic, and infrasound. Its data are analysed at an International Data Centre (IDC) using data contributed by the radionuclide, seismic, infrasound, and hydroacoustic networks, which is in Vienna.

Each method has a different role. Seismic networks are the main tool for underground events, and seismic methods can detect and identify explosions down to or below a yield of about 1 kiloton worldwide. Radionuclide sampling is used to confirm that a seismic event was nuclear. The radionuclide network exists to verify the nuclear nature of an event detected and located by the other technologies. For atmospheric tests, nuclear tests in the atmosphere are best monitored with radionuclide, infrasound, and electromagnetic systems. Hydroacoustic stations, which listen for underwater signals, are also used; signals from the 2017 North Korean nuclear test were recorded on both seismic and hydrophone data. Analysts still face limits, because the structure of the propagation medium is only partly known, which complicates location and yield estimates.

Compensation, relocation and cleanup

The human and environmental costs of testing have been addressed unevenly. On Enewetak, residents were relocated to Rongelap, Utirik, and Ailinginae Atolls, but they were all exposed to nuclear fallout. Return to Rongelap was delayed for a long period because of unsafe conditions, which the U.S. government's own records document. The most visible remnant is the Runit Dome on Enewetak, a hastily constructed concrete lid built over a hole in the ground that was backfilled with radioactive waste. Reports indicate that the Runit dome is reported to be leeching radioactive material into the surrounding sand, which has raised concerns about its long-term integrity.

Beyond the Marshall Islands, compensation programmes have been established in several countries, including the Radiation Exposure Compensation Act of 1990 in the United States for certain downwinders and uranium workers. Kazakhstan closed the Semipalatinsk site in 1991 and has pursued rehabilitation of the affected land and population since then. Cleanup at the Nevada and Pacific sites is ongoing, and disputes over the adequacy and scope of compensation remain unresolved in several places.

Arms control and disarmament treaties

Arms control began with limits on testing before it reached arsenals. The Partial Test Ban Treaty, concluded in Moscow in August 1963, prohibited nuclear weapons tests in three environments: the atmosphere, outer space, and underwater. It left underground testing open, provided such tests did not release radioactive debris beyond the borders of the testing state, because the powers could not agree on verification methods for underground explosions.

The Non-Proliferation Treaty

The Treaty on the Non-Proliferation of Nuclear Weapons grew out of fears that the risk of nuclear war would increase as more countries acquired nuclear weapons. It was opened for signature in 1968 and entered into force on March 5, 1970. Its architecture rests on three pillars: non-proliferation, peaceful use of nuclear energy, and disarmament. In practice, all other signatories are understood to be non-nuclear states and pledge not to acquire nuclear weapons, while the nuclear weapon states commit not to transfer nuclear weapons to any non-nuclear state. In return, the nuclear-armed parties accept a duty to pursue disarmament. India, Pakistan and Israel never joined the treaty, and North Korea withdrew from it in 2003.

SALT, START and New START

Bilateral US–Soviet limits began with the SALT I talks in 1969. In 1981, the United States and the Soviet Union agreed to abide by the unratified 1979 SALT II Treaty, an informal precedent that later returned in debate. The START I agreement was signed in 1991. The most recent pact, New START, was signed in April 2010 and entered into force in February 2011. It set no more than 700 deployed intercontinental ballistic missiles, submarine-launched ballistic missiles and heavy bombers equipped for nuclear weapons; no more than 800 deployed and non-deployed launchers; and no more than 1550 deployed strategic warheads. Its verification regime relied on on-site inspections and data exchanges designed to ensure compliance. Washington and Moscow extended the treaty for five years in 2021, but in February 2023, Russian officials said Moscow would observe the numerical limits but would suspend the treaty's monitoring measures.

Expiry on 5 February 2026

New START ran out on 5 February 2026, though some sources give 4 or 6 February. The treaty expired without a U.S. response to a Russian offer to continue informally abiding by the central limits of the agreement. For the first time in more than half a century, there are no constraints on U.S. and Russian deployment of strategic nuclear weapons. The treaty also had gaps: it did not cover nonstrategic or "theater" nuclear weapons and imposed no limits on China.

Whether restraint will continue informally is disputed. In September 2025 Putin proposed observing the central limits for one more year, and Trump called for experts to work on a new, improved, and modernized Treaty that can last long into the future. Axios reported that negotiators discussed a possible informal deal to observe the pact's limits for at least six months, but the Kremlin said any such extension could only be formal. Russian Foreign Minister Lavrov said Moscow would keep to the limits only as long as the United States does not exceed the above limits. A senior US State Department official said on 17 February that he knew of no such agreement, and he described the treaty's omissions as fundamental flaws.

The CTBT, IAEA safeguards and the TPNW

The Comprehensive Nuclear-Test-Ban Treaty was adopted by the United Nations General Assembly in 1996 but is not in force, by which states agree to ban all nuclear explosions in all environments. Its entry into force requires ratification by a list of named states whose participation is considered necessary, and several of them have not ratified, so the treaty remains unenforced as a legal regime, though an international monitoring network operates regardless.

Verification of civilian nuclear material under the NPT falls to the International Atomic Energy Agency, which applies safeguards inspections to declared facilities. These cover reactors and fuel cycles and are distinct from the arms-control verification of warheads and launchers.

The Treaty on the Prohibition of Nuclear Weapons was adopted in 2017 and entered into force in January 2021. None of the nine nuclear-armed states has joined it, and its relationship to the NPT remains contested.

Global arsenals today

The Stockholm International Peace Research Institute (SIPRI) put the world's inventory at about 12,187 nuclear warheads in January 2026. Roughly 9,745 of them were held in military stockpiles for potential use. The remainder are retired warheads that are waiting to be dismantled, so the stockpile figure is the more useful measure of active forces.

Of the stockpiled warheads, SIPRI estimates that 4,012 were deployed with missiles and aircraft, and the rest were kept in central storage. Between 2,100 and 2,200 of the deployed warheads were held at high operational alert on ballistic missiles, which means they can be launched within minutes. Nearly all of these alert warheads belong to Russia and the United States, with a smaller share held by France and the United Kingdom. SIPRI also judges that China and India may now occasionally put a small number of warheads on missiles during peacetime. For China, the estimate is perhaps as many as 34 warheads with operational forces.

Russia and the United States remain by far the largest holders. SIPRI's table gives the United States about 5,042 warheads in total, of which about 3,700 are in the stockpile and about 1,770 are deployed. Russia's total is about 5,420, with about 4,400 in the stockpile and about 1,796 deployed. Both countries also hold more than 1,000 retired warheads each, which they are gradually dismantling. SIPRI notes that the global total is falling only because of these dismantlements, and that the annual pace of dismantlement appears to be slowing. Both states have extensive modernization programmes, and SIPRI expects their deployed forces to grow in the years ahead.

The other nuclear-armed states are much smaller but several are expanding. SIPRI's table lists about 620 warheads for China, which it describes as the fastest-growing arsenal, with hundreds of missiles loaded into silo fields in the north and further silos under construction in the east. India is estimated at about 190 warheads and Pakistan at about 170, with India's 12 deployed warheads representing a new development. The United Kingdom is estimated at about 225 warheads and France at about 370, of which about 290 are in the stockpile. North Korea is thought to have assembled around 60 warheads and to possess enough fissile material for at least 30 more. Israel is also counted among the nine nuclear-armed states, although it does not publicly acknowledge its arsenal.

All of these numbers are estimates, and the uncertainty is considerable. Governments publish little detail about their stockpiles, so independent researchers rely on open sources, declared treaty data and official statements, and they must infer the status of warheads that are in storage or awaiting dismantlement. Estimates can therefore differ. In March 2026, for example, the commander of United States Strategic Command cited approximately 4,600 Russian warheads, a figure that does not map neatly onto SIPRI's categories of 4,400 in the stockpile and 5,420 in total. The distinction between deployed, stored and retired warheads also matters for statistics of this kind, since a single headline total can combine very different levels of readiness. The arsenal figures are further shaped by the cold-war legacy, in which the largest stockpiles were built up and are now being reduced far more slowly than before.

Deterrence doctrine and strategic stability

Mutually assured destruction and second-strike capability

Deterrence between nuclear powers depends on the premise that a first strike cannot eliminate the other side's ability to retaliate. A state therefore needs second-strike capability, meaning surviving forces that can still strike back after an enemy's surprise attack. Mutually assured destruction (MAD) describes the situation in which both sides hold such forces, so that any attack would bring retaliation sufficient to destroy the attacker as a society. The doctrine is linked with John von Neumann among its early proponents. In the early 1960s, U.S. Secretary of Defense Robert McNamara adopted "assured destruction" as a planning standard: the ability to destroy a large share of Soviet population and industry even after absorbing a surprise attack.

Survivability was the technical core of the idea. Early U.S. airborne alert, in which bombers kept continuous patrols near the Soviet border, offered only a partial guarantee of retaliation. A genuinely survivable force became feasible with ballistic missile submarines, beginning with the George Washington class commissioned in 1959, and later with hardened silos for land-based missiles. By the late 1960s Soviet strategic forces had approached parity with those of the United States, and both superpowers considered their second-strike forces secure against a disarming blow.

MAD also produced counterintuitive policy conclusions. A large-scale civil shelter programme on one side would reduce that side's vulnerability and therefore weaken the other side's assurance of retaliation, so it was seen as destabilizing; for similar reasons, missile defences were judged dangerous because they might appear to blunt a retaliatory strike. Advocates argued that in a crisis neither side could gain by striking first, since no state could limit its own losses. Critics replied that the doctrine did not remove the danger of crisis instability, and that its logic would be disastrous if deterrence failed.

Extended deterrence and the nuclear umbrella

Most nuclear security commitments extend beyond a state's own territory. Extended deterrence means that a nuclear power threatens retaliation in order to protect allies or states regarded as vital interests against attack by a third party. Its main problem is credibility. Against an adversary with a secure second strike, the protecting power must be prepared to expose its own cities to protect another country. During the Cold War, the United States threatened nuclear use in response to a Soviet invasion of Western Europe, and the question of whether Washington would actually carry out that threat was a constant concern for European governments.

Alliances have tried to make such guarantees more concrete through nuclear sharing. NATO's arrangements combine nuclear capabilities, aircraft and infrastructure supplied by several member states; the United States has deployed a limited number of B61 bombs at sites in Europe, and the Nuclear Planning Group provides a forum for consultation and political control over the alliance's nuclear mission. The arrangements are designed to share political responsibility and decision-making rather than the weapons themselves. Recent years have added new layers: a July 2025 declaration between the United Kingdom and France pledged deeper nuclear cooperation, and in March 2026 French President Emmanuel Macron announced a "forward deterrence" strategy that includes cooperation with neighbouring states. European debate has been shaped by doubts about the reliability of U.S. commitments, and by proposals for a European nuclear pillar that would not depend solely on Washington.

No-first-use pledges and declaratory policy

Scholars distinguish declaratory policy, which is what a government says about when it would use nuclear weapons, from operational policy, which is what its forces are actually built and ready to do. A no-first-use (NFU) pledge is a declaratory commitment not to use nuclear weapons unless the state has first been attacked with them. Analysts often note that NFU has been rejected by some nuclear-armed states and accepted only at the declaratory level by most others. Genuine adoption would require changes to the purposes for which weapons are deployed and to the way forces are postured, not merely a statement.

China is the clearest case of a long-standing NFU stance. Its 1975 combat regulations stated that it would never use nuclear weapons first, and it has repeated this position in its 2019 and 2023 defence white papers. Beijing has also promoted NFU treaty proposals among the nuclear-weapon states since 1994, and in 2024 it submitted a further initiative to the NPT review process. Other nuclear states have been largely unwilling to take such proposals forward, and some analysts argue that China uses them to divert attention from its own arsenal growth.

In the United States, the debate is long-running. Proponents argue that Washington already acts as though it had an NFU policy and that its conventional superiority can deter non-nuclear threats. Critics contend that allies in Europe and East Asia would not accept a unilateral pledge, which might encourage adversaries to attack with conventional, chemical, biological or cyber means. The question of whether nuclear weapons should be restricted to responses to existential threats sits within the same debate, and it remains unresolved.

Game-theoretic analysis and the risk of miscalculation

Deterrence is often modelled with game theory, in which the value of an action depends on how an opponent will respond. Under mutual second-strike capability, the outcome in which neither side attacks behaves like a Nash equilibrium: no side can gain by deviating unilaterally. Yet the structure also resembles a prisoner's dilemma. Each side might prefer mutual restraint, but fear that the other will exploit any reduction in its own forces, which drives continued arms accumulation.

These models rest on strong assumptions. They presume that leaders are rational, value their own survival, and calculate costs and benefits accurately. In practice, decision theory points to bounded rationality and to information asymmetry: leaders cannot know an adversary's intentions, the reliability of its command systems, or whether a second-strike force will actually survive. Misreading any of these can lead to a belief that a first strike is safe, or that a threat is bluff. Time pressure during a crisis reduces the room for reflection, and accidents or false warnings can be interpreted as real attacks. Recent analyses have argued that combining verified reductions, NFU pledges and investment in early-warning systems can lower these risks, while survey data suggest that the norm against nuclear use remains widely held even among people who accept its use in extreme circumstances.

Command and control, safety and accidents

Chains of authority and time pressure

Nuclear forces are run through layered chains of command, in which political leaders authorize use and military units execute the order. Some states have adopted launch-on-warning postures, under which missiles could be fired after sensors register an incoming attack but before any warhead arrives. Such postures shrink the decision window to minutes and make the outcome depend on the reliability of sensors, communications and the people who interpret their reports. These pressures are a central concern of decision theory and of analysts who study bounded rationality under stress.

Engineering safeguards

Designers use several measures to prevent accidental or unauthorized detonation. Permissive action links are electronic locks that require a coded authorization before a weapon can be armed, and they were developed from the early 1960s. One-point safety is a principle under which an accidental detonation of the high explosives at any single point should not produce a full nuclear yield, because the implosion would not be symmetrical. Insensitive high explosives are formulations that are hard to set off by impact, fire or bullets, so a crash or fire is less likely to trigger detonation. Mechanical arming sequences add further steps that normally cannot be completed by accident.

Documented accidents

On 24 January 1961, a B-52 carrying two Mark 39 thermonuclear bombs broke apart over Goldsboro, North Carolina. Both bombs separated from the aircraft and began their firing sequences, despite safeguards meant to prevent this. One weapon's parachute opened and the bomb reached the arm/safe switch stage, where the safety mechanism held. Engineers later judged that this weapon was only one switch away from detonation. The second bomb buried itself deep in soft ground, and investigators examined whether a one-point detonation had occurred before concluding it had not. Estimates of the yield of these weapons vary, and information about the accident was declassified only in part, with fuller details released in 2013.

On 17 January 1966, a B-52G collided with a KC-135 tanker during refuelling over the Mediterranean off Spain. The bomber carried four Mark 28 thermonuclear weapons. In the crash, the conventional explosives in two of the weapons detonated on impact, scattering plutonium over roughly 2 square kilometres near Palomares. None of the weapons produced a nuclear yield. The fourth bomb fell into the sea and was recovered from about 2,850 feet of water roughly two and a half months later. Afterwards, nuclear-armed flights over Spanish territory were prohibited.

On 26 September 1983, the Soviet Oko satellite early-warning system reported a launch of US intercontinental missiles. Lieutenant Colonel Stanislav Petrov, the duty officer at the Serpukhov-15 command centre near Moscow, judged the alert to be false, reasoning that a genuine first strike would involve many more missiles than the one to five reported. He did not pass the alarm up as a confirmed attack. Investigators later attributed the warning to sunlight reflected from high-altitude clouds. Accounts differ on exactly what he reported to his superiors, but his judgement is widely regarded as having prevented a retaliatory strike.

Lessons for alert levels and system design

These incidents point to several recurring lessons. Warnings from a single sensor or computer should not be enough to trigger a response, and independent confirmation from separate systems, such as radar and ground-based detection, is needed before escalation. Alert levels and procedures must allow time for verification, and communication channels between adversaries should remain reliable during a crisis. Software and sensor errors must be anticipated, tested and corrected, since a system that works exactly as designed can still produce a false conclusion. Finally, the accidents show that physical safeguards, such as one-point safety and insensitive explosives, reduce the risk of catastrophe but cannot replace careful human judgement and clear procedures.

Health and epidemiological consequences

Acute radiation syndrome

Acute radiation syndrome (ARS), also called radiation sickness, follows a high dose of penetrating radiation delivered to much of the body within minutes. The gray (Gy) measures absorbed dose: one gray equals one joule of ionizing energy deposited per kilogram of tissue. The dose needed is large. The lowest threshold for ARS is a dose of greater than about 0.7 Gy, although mild symptoms may appear at doses as low as 0.3 Gy. The exposure is usually external, because radioactive material deposited inside the body has caused ARS only in extremely rare cases.

Severity depends on dose and produces distinct syndromes. The bone marrow (hematopoietic) form is the most common, and its primary cause of death is destruction of the bone marrow, leading to infection and hemorrhage. At higher doses, the gastrointestinal syndrome occurs above 10 Gy and is associated with mostly fatal infection, dehydration, and electrolyte imbalance. Above about 50 Gy, a cardiovascular and central nervous system syndrome appears, with death within days. Lethality is also time-dependent: about half of patients exposed to roughly 3 to 4 Gy may die without prompt medical support, and supportive care raises that threshold substantially. Lower dose rates also raise the thresholds, since the same dose spread over a longer period is less likely to cause ARS.

Long-term cancer risk and the Life Span Study

The principal quantitative source for radiation-induced cancer risk is the Life Span Study (LSS) of atomic-bomb survivors. Its cohort of 120,321 subjects includes 93,741 atomic bomb survivors of Hiroshima and Nagasaki and 26,580 persons who were not in either city at the time of the bombings. Cancer incidence follow-up began in 1958 and was analysed through 2009, with cases identified by linkage to population-based cancer registries (the follow-up period was 1958–2009, providing 3,079,484 person-years of follow-up. Cases were identified by linkage with population-based Hiroshima and Nagasaki Cancer Registries.). The analyses estimate excess relative risk (ERR) and excess absolute risk (EAR) per gray of weighted colon dose, adjusted for smoking. Sixty-four years after exposure, solid cancer risks remain elevated. Leukemia, which rises earlier and peaks sooner than solid tumours because blood-forming tissue is highly radiosensitive, was among the first excess cancers identified in the cohort.

Differences by age, sex and exposure route

Risk depends strongly on age at exposure. Excess absolute rates for solid cancer keep increasing throughout life for all ages at exposure. Solid cancer risks are especially high among people exposed in utero and in early childhood. The pattern varies by organ: for liver cancer, relative risk was significantly raised for exposures at ages 0–29 but not after age 30 (ERRs by age-at-exposure categories were significantly increased among those exposed at 0–9, 10–19 and 20–29 years, but not after age 30).

Sex-specific patterns also appear. For pancreatic cancer, the dose-related increase was statistically significant among women but not among men. Male breast cancer was also found to carry a radiation-related excess. On exposure type, most of the dose to survivors came from gamma radiation (the accuracy of the gamma dose is central to the accuracy of the dosimetry system because most of the dose to survivors is from gamma rays). Residual radioactivity from the bombings was negligible for most LSS participants compared with initial radiation, so the study models mainly prompt exposure. Internal exposure is therefore a separate question that the bombing data address poorly.

Dose reconstruction and study methods

Individual doses are not measured directly and must be reconstructed. The earlier dosimetry system, DS86, was replaced after thermal neutron activation measurements in Hiroshima appeared to show a different distance trend than DS86 estimates, calling its accuracy into question. The new system, DS02, was developed by an independent group of scientists, the Joint U.S.-Japan Working Group. It revised the Hiroshima yield from 15 to 16 kilotons and the burst height from 580 to 600 metres (the working groups decided to increase the estimated height of the nuclear explosion from 580m in DS86 to 600m in DS02, and to increase the estimated bomb yield from 15kt in DS86 to 16kt in DS02.). Gamma doses at distances beyond 1 km, where most survivors were, were about 10% higher than under DS86. Later analyses use an updated version, DS02R1, and the shielding provided by terrain and large buildings was recalculated.

Statistical reporting is essential to interpreting these results. Models use Poisson regression, and results are presented with confidence intervals. For example, the pancreatic cancer excess per gray was 0.38, with a 95% confidence interval from below zero to 0.83, which was not statistically significant. Dose estimates for accident and test populations are reconstructed by different methods, and clinical ARS cases among such groups, such as the firefighters that first responded after the Chernobyl Nuclear Power Plant event in 1986, are assessed separately from the long-term cohort.

Ethics, international law and the legal status of use

Humanitarian law as the legal test

Legal analysis of nuclear use relies on the law of armed conflict, which is also called international humanitarian law. Three principles do most of the work. Distinction bars weapons that cannot separate military targets from civilians. Proportionality forbids attacks whose likely harm to civilians would exceed the military advantage sought. Necessity allows only the force needed to reach a legitimate military aim. Distinction prohibits a weapon that cannot discriminate in its effects between military and civilian targets. Proportionality prohibits a weapon whose probable effects on noncombatants would likely be disproportionate to a specific, legitimate military objective. Necessity permits only the level of force needed to achieve the military objective.

Applying these rules to nuclear weapons is contested. The 1996 ICJ opinion itself stressed that these weapons release immense heat and energy along with prolonged radiation, which places them in a class of their own. The Court described their key characteristics as the enormous heat and energy they produce and the powerful, long-lasting radiation they release, making them "potentially catastrophic." In 2011 the International Red Cross and Red Crescent Movement's Council of Delegates concluded that it was hard to see how any use could comply with distinction, precaution and proportionality. Governments that possess these weapons reject the view that their use is inherently unlawful; the United States has stated that any nuclear operations would be conducted in line with the law of armed conflict, and that civilian damage would be minimised as far as objectives allow.

The 1996 advisory opinion

On 8 July 1996 the International Court of Justice, asked by the UN General Assembly, found that the threat or use of nuclear weapons would generally be contrary to international law applicable to armed conflict, particularly humanitarian law. The Court, however, could not conclude definitively whether such a threat or use would be lawful or unlawful in an extreme circumstance of self-defence in which a state's very survival was at stake. Its most consequential finding was unanimous: there exists an obligation to pursue in good faith and bring to a conclusion negotiations leading to nuclear disarmament in all its aspects under strict and effective international control. That conclusion has since been treated as the authoritative reading of Article VI of the Nuclear Non-Proliferation Treaty, and in 2002 a General Assembly vote endorsing it passed 161 to 4 with one abstention.

The opinion was not unanimous in its reasoning. Several judges dissented, and critics argue that the Court's inability to rule on the self-defence case leaves a significant gap. Others note that the Court did not say a nuclear response could never be proportionate, but stressed that the particular risks of these weapons must weigh heavily in any such assessment.

Ethics of deterrence and moral responsibility

The ethical debate divides along a familiar line. Critics argue that a policy built on the credible threat of mass civilian casualties is morally indefensible, whether or not the threat is ever carried out, and that intending to do what international law forbids cannot be squared with that law. Defenders argue that deterrence has helped prevent direct war between great powers for more than seven decades, and that the moral question is whether a threat can be credible and restrained without requiring the actual use of force. Both sides accept that states which possess these weapons carry particular responsibilities: to keep them under strict command and control, to avoid accidental use, and, for the nuclear-armed states that are parties to the NPT, to pursue disarmament in good faith.

The Treaty on the Prohibition of Nuclear Weapons

The 2017 Treaty on the Prohibition of Nuclear Weapons (TPNW) was adopted by a vote of 122-1-1 after the second and final round of negotiations in New York. The Netherlands voted against adoption, and Singapore abstained. Its prohibitions cover development, testing, production, possession, stockpiling, use and threat of use of nuclear weapons, as well as stationing another country's weapons on a party's territory. After the 50th instrument of ratification was deposited on 24 October 2020, the treaty entered into force on 22 January 2021. The treaty also requires parties to provide adequate assistance to people affected by the use or testing of nuclear weapons, and to carry out environmental remediation in areas under their jurisdiction.

Its reach is limited. It does not apply to states that have not joined it. None of the nuclear-armed states, nor any of the allied states that rely on extended nuclear deterrence, have supported, signed or ratified it. NATO has rejected the treaty explicitly. In its 2021 summit communiqué the alliance restated its opposition, arguing that the treaty conflicts with its deterrence policy and risks undermining the NPT. Supporters respond that the treaty complements existing non-proliferation and disarmament instruments, and that it strengthens the NPT's disarmament commitments. Its opponents, by contrast, argue that it is divisive and could weaken the existing framework. The treaty also contains a route for states that possess weapons to eliminate them, which Article 4 sets out as a process for eliminating stockpiles.

Modernization and emerging technologies

Current modernization programmes

Most of the world's nuclear-armed states are replacing or upgrading their forces at the same time. SIPRI reports that all nine nuclear-armed states continued modernization in 2025, and most of them fielded new nuclear-armed or nuclear-capable systems that year (The nine nuclear-armed states continued programmes to modernize and enhance their nuclear arsenals in 2025, and most deployed new nuclear-armed or nuclear-capable weapon systems during the year.).

In the United States, the country is replacing or modernizing nearly every component of its strategic nuclear arsenal. The Sentinel ICBM, which is meant to replace the Minuteman III, has been the most troubled part. Its cost rose from an initial estimate of about $78 billion to more than $141 billion, and its first operating capability slipped by at least two years (Sentinel was designed to replace the aging, 400 deployed Minuteman III ICBMs, but the Sentinel program was halted in 2024 for review after costs rose from an initial $78 billion to over $141 billion). The Pentagon now plans a first flight from a missile pad in 2027 and a Milestone B decision before the end of 2026 (a missile pad launch scheduled for 2027, which will mark a pivotal moment in our flight test campaign. Sentinel is on a path to its Milestone B decision by the end of this year). The Columbia-class submarines are also behind schedule, and the Pentagon is considering three more boats, which would bring the class to 15 (the department is considering purchasing three more Columbia-class submarines, potentially bringing the total class to 15 boats). The B-21 bomber is intended to be dual-capable, and its production rate has been raised by 25 percent (a 25 percent increase in B-21 Raider production). A bomber force structure study is also reviewing how many B-21s to buy and how the B-52H will be replaced (a bomber force structure study that will re-consider how many of the new B-21 Raider bombers are purchased and how the B-52H heavy bomber will be replaced).

Russia's programme has a long record of announcements running ahead of performance. On 12 May 2026 Russia reported a test of the RS-28 Sarmat, which was fired from Plesetsk toward a target about 5,700 km away (The missile was fired from a test silo at the Plesetsk Cosmodrome in northern Russia, reportedly hitting its target at the Kura test range about 5,700 kilometers away.). Analysts note that this was only the first successful flight since 2022 and that the missile had suffered years of delays and failed tests (It is the first known successful flight test since April 2022.). Putin also said the nuclear-armed Poseidon underwater drone and the Burevestnik cruise missile were in their "final stages" of development. Independent assessments are more cautious. One analysis judges that the two systems are unlikely to achieve operational status in the next 5 years, and an MIT calculation put Burevestnik's speed at about Mach 0.75, which contradicts Russian claims of a fast, defence-evading weapon (MIT researchers Hecla and Kemp calculated that Burevestnik flies at approximately Mach 0.75 and measures approximately 9.5 metres in length. These parameters are inconsistent with Russian claims of a fast, defence-evading weapon.). After New START expired in February 2026, one analysis expects Russia to add warheads to existing delivery systems (Russia will likely increase the number of warheads uploaded onto existing strategic delivery systems following the expiration of New START.).

China has the most rapid expansion. SIPRI reports that by January 2026, China had loaded hundreds of missiles into three large missile silo fields in the north of the country, while working to complete 30 silos in three mountainous areas in the east. Its ICBM force could reach parity with the other two major arsenals by the end of the decade, though a Bulletin of the Atomic Scientists assessment finds no evidence that China's ongoing nuclear expansion will result in parity with the US arsenal. The United Kingdom is replacing its Vanguard-class submarines with Dreadnought-class boats, France is developing a successor submarine with the M51 missile family and a new air-launched weapon for its Rafale aircraft, and India is developing its Arihant-class submarines, the K-4 submarine-launched missile and the Agni-V with multiple warheads. These programmes are not covered in detail by the sources used here and are described from general background knowledge.

Hypersonic glide vehicles and missile defence

Hypersonic glide vehicles are launched on a ballistic missile and then manoeuvre through the atmosphere at very high speed. Russia's Avangard, which was presented in 2018, is capable of flying 27 times faster than the speed of sound. The Oreshnik, which is widely assessed as a rebadged RS-26 Rubezh produced in limited numbers, has been used in combat. Because these vehicles fly lower and change course, they can shorten warning time and make tracking harder for radars and interceptors.

Russian officials frame many of these weapons as a response to missile defence. Putin has said the new systems answer the missile shield that Washington built after it left the 2002 ABM Treaty (Putin has described those new weapons as part of Russia's response to the US missile shield that Washington developed after its 2001 withdrawal from a Cold War-era US-Soviet Union pact that limited missile defences.). The underlying dynamic is that defences designed against a small number of attacks may undermine a rival's confidence in its second-strike force, which prompts offensive upgrades and more warheads. This action-reaction cycle is the central strategic-stability concern, and it is why interceptors, manoeuvring warheads and multiple warheads are assessed together.

Artificial intelligence, cyber operations and space sensors

Emerging digital technologies are being applied to early warning, target identification, communications and decision support. Sensor networks in space can detect launches, while artificial intelligence can help sort large volumes of sensor data and suggest options to commanders. US defence officials have cited artificial intelligence, autonomy as reasons to keep investing in modernization, although the full policy rationale is not described in the sources used here. Cyber operations add a further layer: networks that link sensors, command posts and missile fields can be attacked, spoofed or degraded, and a compromise of warning data could mislead decision-makers.

Risks of miscalculation

The main concern is that these technologies compress the time available for decisions. A hypersonic or manoeuvring warhead gives less time to determine what is happening, and an AI-assisted system may present a false reading with the same confidence as a true one. Dual-capable systems, such as missiles or aircraft that can carry either conventional or nuclear warheads, can also make it hard for an opponent to know what is being launched. Analysts therefore worry about errors in sensors or software, misread intentions and the pressure to respond before a situation is confirmed. The 1983 false alarm, discussed in the command and control section, shows how computer and sensor errors can already produce serious alerts. Reducing these risks depends on the same tools that create them, which is why arms control and verification of new systems are as important as the technology itself.

Relationship to civil nuclear technology

Reactors and weapons: a difference of design and degree

Civil reactors and weapons share the same physics of fission, but they are built to do opposite things with it. The most common commercial design, the light-water reactor, runs on fuel that is only slightly enriched, and ordinary water slows neutrons so that fission proceeds in a controlled way. LWRs use LEU (usually between 3 and 5 percent enriched) as nuclear fuel in the reactor core and natural water (light water) to moderate (slow down) neutrons for increased fission efficiency. A weapon is designed for the reverse: it needs a fast, uncontrolled chain reaction, which releases a large amount of energy compared with the other isotopes. Reactor fuel is also consumed in a way that spreads its products across a long operating cycle, whereas a weapon's core is machined, assembled and detonated once. These differences in design mean that a reactor cannot be converted into a weapon by simply changing the fuel. The conversion requires separate facilities for enrichment or reprocessing, which is why the proliferation debate centres on those stages rather than on reactors as such.

Enrichment levels and the LEU–HEU boundary

Natural uranium contains only 0.7% of the chain-reacting isotope uranium-235 (U-235). Standard terminology divides enriched material into categories. One widely used classification places Low-Enriched Uranium (LEU) between 0.72 and 20 percent U-235 and Highly Enriched Uranium (HEU) from 20.00 to 89.99 percent U-235, with weapons-grade material at 90 percent or above. The 20 percent threshold is partly conventional. Analysts note that material below 90 percent is not useless for weapons; it simply requires more material and more effort. One handbook states that HEU enriched to the highest levels could be usable, but would be much less efficient than weapons-grade HEU. The cost of this effort is large. A reference table of separative work units shows that producing 1 kg of 90 percent material from natural feed requires about 180 kg of natural uranium and 225 SWU, whereas producing a 4 percent reactor product needs only 7.6 kg of natural uranium and 6.6 SWU. The gap between reactor fuel and weapon material is therefore a matter of scale and effort, not of kind.

Research reactors complicate this picture. Some research reactors use 90% enriched U-235 to produce medical isotopes, so there are civilian applications for this fuel too. Debate over high-assay LEU (HALEU), enriched to roughly 20 percent, has grown for the same reason. Despite a widespread belief that high-assay low-enriched uranium (HALEU) cannot be used directly in a nuclear weapon, independent analysis strongly suggests that it is weapon-usable, although the required quantities are larger than for weapons-grade HEU.

Safeguards and dual-use technology

Nuclear technology is dual-use in the sense that the same equipment and materials serve civil and military ends. Dual-use technologies are those technologies that are designed or suitable for both civilian and military, terrorism, or weapons of mass destruction-related applications. The main legal response is the safeguards system. Under the Non-Proliferation Treaty, all non-nuclear-weapon states must accept IAEA safeguards on their nuclear activities. Safeguards are chiefly concerned with detecting the diversion of nuclear material (plutonium, enriched uranium, or uranium-233) from declared facilities, and they rely on material accounting. The limits of this system are real. Verification depends on sampling, and inspectors can provide only statistical, and not absolute, confidence in nondiversion. Export controls supplement safeguards. The supplier guidelines distinguish a trigger list covering source material, special fissionable material and reactors from a second list of dual-use items that could make a major contribution to a nuclear explosive device or an unsafeguarded nuclear fuel cycle activity. Even deuterium is treated differently depending on its end use, since heavy water -- water enriched in deuterium -- can be used to breed special fissionable material.

Peaceful nuclear explosions

In the 1950s and 1960s, both superpowers explored using nuclear explosives for civil engineering. In the United States, Project Plowshare was the overall United States program for the development of techniques to use nuclear explosives and large non-nuclear explosions for peaceful construction purposes. 35 nuclear warheads were detonated in 27 separate tests. The Soviet programme was larger, with 124 tests. Proposals included a harbour in Alaska and a second canal through Central America. The results were poor. The 1962 Sedan cratering test produced large amounts of fallout, and the 1967 Gasbuggy test still contains nuclear contamination from a single subsurface blast in 1967. Dates for the end of the US programme vary between sources: one places its end in 1975, while another describes termination in 1977 due in large part to public opposition, after $770 million had been spent on the project.

The main reasons for abandonment were practical and political. Projects faced fears of radioactive venting and the cost ineffectiveness of these projects, and the 1963 Limited Test Ban Treaty made underground releases of debris across borders a legal risk. A national environmental movement and rising public opposition to nuclear applications also weighed heavily. Critics had long suspected that civil programmes were a cover for weapons testing, and that suspicion, combined with the mounting environmental costs, made the case for peaceful explosions untenable.

Nuclear power and the proliferation debate

Nuclear power remains a major source of electricity, and this is the main reason the civil–military boundary is contested. At low enrichment levels, uranium powers nearly 10% of the world's electricity. Because every enrichment plant could in principle produce weapon material, states that want reactors also want the technology needed to make fuel. The NPT is built around this tension. Non-nuclear states agree not to acquire weapons, while nuclear states commit to disarmament and to sharing peaceful technology, which is conditional on safeguards compliance. The trade-off intensifies as fuel designs change. Analysis of small modular reactor fuel has found that the amount of work needed to enrich up to 95% for the case of 20% LEU is nearly three times easier when compared to the work done to enrich to 95% from 5% LEU. Illicit transfers have also shown the risk: the illicit transfer of enrichment technology from Pakistan to countries like the Democratic People's Republic of Korea (North Korea) and Iran has underscored the urgency of stringent measures. The policy debate thus turns on whether civil nuclear power can be expanded while keeping enrichment and reprocessing within verifiable limits.

References

1. The Binding Energy Curvevarsitytutors.com
2. Prompt criticalityen.wikipedia.org
3. Deuterium%E2%80%93tritium fusionen.wikipedia.org
4. Ignition temperature - Glossary - Energy Encyclopediaenergyencyclopedia.com
5. Iron-56en.wikipedia.org
6. Prompt neutronen.wikipedia.org
7. Delayed Neutron Precursor - an overviewsciencedirect.com
8. Westinghouse Technology Systems Manual Section 2.1 Reactor Physics Reviewnrc.gov
9. Substantially enhanced deuteron-triton fusion probabilities in intense low-frequency laser fieldsarxiv.org
10. Neutrinos from core-collapse supernovaearxiv.org
11. Little Boy: A Gun-Type Bombatomicarchive.com
12. Little Boyen.wikipedia.org
13. Fat Man and Little Boy: The Design of the Atomic Bombslflank.wordpress.com
14. Designs of Two Bombsatomicarchive.com
15. Gun-type fission weaponen.wikipedia.org
16. Nuclear Weapons Primerwisconsinproject.org
17. Nuclear weapon - Gun Assembly, Implosion, Boostingbritannica.com
18. Early nuclear weaponsnuclearweaponsedproj.mit.edu
19. Electronics and Detonators - Nuclear Museumahf.nuclearmuseum.org
20. Boosted fission weaponen.wikipedia.org
21. Introduction to Nuclear Weapon Physics and Designnuclearweaponarchive.org
22. Thermonuclear weaponen.wikipedia.org
23. Radiation implosionen.wikipedia.org
24. Thermonuclear weaponmilitary-history.fandom.com
25. History of the Teller%E2%80%93Ulam designen.wikipedia.org
26. Operation Ivyen.wikipedia.org
27. Bravo thermonuclear test - Studentskids.britannica.com
28. Castle Bravoen.wikipedia.org
29. RDS-37en.wikipedia.org
30. History of the Teller–Ulam designmilitary-history.fandom.com
31. Big Ivan, The Tsar Bomba (“King of Bombs”)nuclearweaponarchive.org
32. Tsar Bombaen.wikipedia.org
33. Nuclear weapon yielden.wikipedia.org
34. Tsar Bomba History, Test & Consequencesstudy.com
35. Operation Greenhouseen.wikipedia.org
36. Full article: The Teller–Ulam Innovation at 75: An Idea that Changed the World and Disputes over Credittandfonline.com
37. Sundial (weapon)en.wikipedia.org
38. Neutron bombmilitary-history.fandom.com
39. W70en.wikipedia.org
40. W70globalsecurity.org
41. The Neutron Bombairandspaceforces.com
42. The United States Announces Production of Neutron Bombsebsco.com
43. W79 Artillery-Fired Atomic Projectileen.wikipedia.org
44. W76-2: Low-Yield Nuclear Warheadmodeldiplomat.com
45. US Deploys New Low-Yield Nuclear Submarine Warhead - Federation of American Scientistsfas.org
46. Pentagon Confirms Low-yield SLBM Deploymentmissilethreat.csis.org
47. US Navy deploys low-yield Trident W76-2 nuclear submarine warheadnaval-technology.com
48. US Navy’s Deployment of New “Low-Yield” Nuclear Warhead a Dangerous Step via The Guardian (Portside)lucian.uchicago.edu
49. smith criticizes the administration s deployment of w76 2 low yield nuclear warheadsdemocrats-armedservices.house.gov
50. Dirty bomben.wikipedia.org
51. Dirty Bombsiwaste.epa.gov
52. A Dirty Bomb Is Not a Nuclear Bomb - Union of Concerned Scientistsblog.ucs.org
53. Backgrounder: Dirty Bombs.nrc.gov
54. Thermal Radiationatomicarchive.com
55. The Devastating Effects of Nuclear Weaponsthereader.mitpress.mit.edu
56. Inaccurate Prediction of Nuclear Weapons Effects and Possible Adverse Influences on Nuclear Terrorism Preparedness - Homeland Security Affairshsaj.org
57. FM 8-9 Part I/Chptr 3 Effects of Nuclear Explosionsnuke.fas.org
58. Cheap Method for Shielding a City from Rocket and Nuclear Warhead Impactsarxiv.org
59. Thermal Radiation Calculatornuclearweaponsedproj.mit.edu
60. Nuclear Detonation: Weapons, Improvised Nuclear Devices - Radiation Emergency Medical Managementremm.hhs.gov
61. Nuclear falloutfallout.fandom.com
62. Nuclear fallouten.wikipedia.org
63. The Effects of Nuclear Weapons - Glasstone and Dolanatomicarchive.com
64. Effects of Nuclear Explosionsnuclearweaponarchive.org
65. NMHB 2020 [Revised] - Chapter 3acq.osd.mil
66. Nuclear Delivery Systemsniaw.org
67. US Nuclear Triad Overview - Stanfordlarge.stanford.edu
68. Rethinking the US Nuclear Triad - Air Universityairuniversity.af.edu
69. Fact Sheet: Multiple Independently-targetable Reentry Vehicle (MIRV) - Center for Arms Control and Non-Proliferationarmscontrolcenter.org
70. Multiple Independently Targetable Re-entry Vehicle (MIRV) » UnderStand UPSCunderstandupsc.com
71. Sage Reference - Encyclopedia of United States National Security - Multiple Independently Targetable Reentry Vehicles (MIRVs)sk.sagepub.com
72. U.S. Strategic Nuclear Forces: Background, Developments, and Issues - EveryCRSReport.comeverycrsreport.com
73. A brief history of the nuclear triadblog.nuclearsecrecy.com
74. Full article: Tactical nuclear weapons, 2019tandfonline.com
75. Tactical Nuclear Weapons Explained: Yield, Delivery & Escalation Risk — MissileStrikes.commissilestrikes.com
76. Nuclear artilleryen.wikipedia.org
77. Tactical ballistic missileen.wikipedia.org
78. Ending Tactical Nuclear Weapons: A Brief History and a Path Forward - The Council on Strategic Riskscouncilonstrategicrisks.org
79. Nuclear triaden.wikipedia.org
80. Multiple independently targetable reentry vehicleen.wikipedia.org
81. Einstein-Szilard Letter - Atomic Heritage Foundationahf.nuclearmuseum.org
82. Albert Einstein's letters to President Franklin Delano Roosevelthypertextbook.com
83. Einstein's Letter to President Roosevelt - 1939opnlttr.com
84. 1939: The Einstein–Szilárd Letter Changed the Nuclear Ageuniversalrecord.org
85. Alexander Sachsen.wikipedia.org
86. Manhattan Projecten.wikipedia.org
87. Manhattan Project (1942–1945): Secret Cities, Trinity Testlibraryofhistory.org
88. Timeline of the Manhattan Projecten.wikipedia.org
89. Manhattan Project: total costs by facility 1945statista.com
90. History of the Hanford Sitelarge.stanford.edu
91. 'Destroyer of Worlds': The Making of an Atomic Bombnationalww2museum.org
92. 80 Years Ago: The First Atomic Explosion, 16 July 1945nsarchive.gwu.edu
93. Manhattan Project: The Trinity Test, July 16, 1945osti.gov
94. Collateral damage: American civilian survivors of the 1945 Trinity test - Bulletin of the Atomic Scientiststhebulletin.org
95. 80 Years- Get the Facts - ICANicanw.org
96. Worst Decisioncfr.org
97. The Day That Changed the World Forever - Center for Arms Control and Non-Proliferationarmscontrolcenter.org
98. Einstein’s Letter to President Roosevelt - 1939atomicarchive.com
99. The atomic bomb: its history and the struggles of scientistsarxiv.org
100. The Lesson of Einstein's Warning to FDR on Nuclear Weaponstime.com
101. Inaugural Soviet Nuclear Test - Multilateral Dialogue Vienna - Konrad-Adenauer-Stiftungkas.de
102. RDS-1en.wikipedia.org
103. RDS-2 Linageen.wikipedia.org
104. Russia and weapons of mass destructionen.wikipedia.org
105. Global nuclear warhead inventory by country 2026| Statistastatista.com
106. Global Nuclear Warhead Stockpiles (1945-2024)visualcapitalist.com
107. World Nuclear Forces 2026: Who Has Nuclear Weapons and How Manyglobalmilitary.net
108. Nuclear Weapons: Who Has What at a Glancearmscontrol.org
109. List of states with nuclear weapons - Wikiwandwikiwand.com
110. Country Nuclear Profilesniaw.org
111. Status of World Nuclear Forces - Federation of American Scientistsfas.org
112. Nuclear risks grow as new arms race looms—new SIPRI Yearbook out nowsipri.org
113. No Promised Land: The Shared Legacy of the Castle Bravo Nuclear Testarmscontrol.org
114. List of military nuclear accidentsen.wikipedia.org
115. Marshall Islands - Atomic Heritage Foundationahf.nuclearmuseum.org
116. 60th Anniversary of Castle BRAVO Nuclear Test, the Worst Nuclear Test in U.S. Historynsarchive2.gwu.edu
117. Those Who Witnessed Castle Bravo Looked Into Armageddonmedium.com
118. graphsearch.epfl.chgraphsearch.epfl.ch
119. Tsar Bomba - Atomic Heritage Foundationahf.nuclearmuseum.org
120. Tsar Bomba and Novaya Zemlya Test Site - Nuclear Blast Simulatornuclearblastsimulator.com
121. 2. CTBT Monitoring Capabilitynap.nationalacademies.org
122. Infrasound Technology Workshop 2019 (ITW2019) (10-November 14, 2019): Contribution List · CTBTO Conferences and Workshops (Indico)conferences.ctbto.org
123. Neutrino Detectors as Tools for Nuclear Securityarxiv.org
124. Full article: The Hydroacoustic Network of the CTBT International Monitoring System: Access and Exploitationtandfonline.com
125. Castle Bravo: Marking the 65th Anniversary of the US Nuclear Disaster - AIIAinternationalaffairs.org.au
126. Castle BRAVO at 70: The Worst Nuclear Test in U.S. Historynsarchive.gwu.edu
127. Tsar Bomba: The Most Powerful Nuclear Weapon Ever Builtscience.howstuffworks.com
128. Tsar Bomba nuclear test, 1961johnstonsarchive.net
129. CONRAD OBSERVATORY CTBTO TEST FACILITY ...ctbto.org
130. PTBT and NPT: Milestones in Nuclear Arms Control • PolSci Institutepolsci.institute
131. Chapter 2 How We’ve Come to View the NPT: Three Pillars Dean Rustnpolicy.org
132. International agreements relating to nuclear weapons - CNDcnduk.org
133. End of New START: Short- and Medium-Term Options - Vienna Center for Disarmament and Non‑Proliferationvcdnp.org
134. New START’s Final Year—What Next?cisac.fsi.stanford.edu
135. Russia says again that it will stick to limits on nuclear weapons in expired New START treaty, if U.S. does - CBS Newscbsnews.com
136. New START Expires As U.S. Urges ‘Modernized’ Treatyarmscontrol.org
137. Sergey Lavrov says Russia will observe nuclear limits after NEW Start expiresthehill.com
138. New START: No 'Gentleman's Agreement' Between US, Russia On Expired Nuclear Treatyrferl.org
139. russia says it will stick to new starts nuclear arms limits as long as the u s doespbs.org
140. Test Ban Treatyen.wikipedia.org
141. New START Expires: What Happens Next?belfercenter.org
142. Russia says it will stick to New START's nuclear arms limits as long as the US doesclickondetroit.com
143. Expiry of the New START Treaty (6 February 2026)diplomatie.gouv.fr
144. Treaty on the Non-Proliferation of Nuclear Weapons2009-2017.state.gov
145. SIPRI Yearbook 2026 Chapter 8. World nuclear forcessipri.org
146. Increasing focus on nuclear weapons amid heightened escalation risks—new SIPRI Yearbook out nowsipri.org
147. Armaments, Disarmament and International Security SIPRI YEARBOOK 2026 Summarysipri.org
148. Nuclear risks rise as powers expand and modernise arsenals: SIPRI studyaljazeera.com
149. SIPRI says nuclear-armed states are expanding and modernising arsenals as escalation risks and arms control pressures growdefence-industry.eu
150. SIPRI Report on Nuclear Weaponsdhyeyaias.com
151. Is India Putting Nuclear Warheads to Sea? What the 2026 SIPRI Yearbook Actually Saysindiasworld.in
152. un 230612 sipri01globalsecurity.org
153. Mutual assured destruction (MAD)britannica.com
154. Second strikeen.wikipedia.org
155. Mutually assured destructionen.wikipedia.org
156. Mutually Assured Destruction (MAD) — Definition & Meaningmodeldiplomat.com
157. An Overview of Mutual Assured Destructionbootcampmilitaryfitnessinstitute.com
158. The Cold Comfort of Mutually Assured Destructionwarontherocks.com
159. Gov 312L Module 15 Flashcardsquizlet.com
160. NATO's Nuclear Sharing Arrangementsnato.int
161. NATO’s Nuclear Sharingdefencetoday.com
162. U.S. Extended Deterrence and Regional Nuclear Capabilitiescongress.gov
163. France should join NATO’s nuclear sharing arrangements to strengthen European deterrencechathamhouse.org
164. The Utility of Nuclear Weapons and the Strategy of No-First-Usebelfercenter.org
165. No first useen.wikipedia.org
166. No-first-use of Nuclear Weapons Initiative_Ministry of Foreign Affairs of the People's Republic of Chinamfa.gov.cn
167. 'No First Use' and Nuclear Weaponscfr.org
168. PONI Live Debate- U.S. No First Usecsis.org
169. The nuclear paradox: Why the world wants both deterrence and disarmament - Bulletin of the Atomic Scientiststhebulletin.org
170. No-First Use of Nuclear Weapons: A Policy Assessment - DTICapps.dtic.mil
171. 1961 Goldsboro B-52 crashen.wikipedia.org
172. Mark 39 nuclear bomben.wikipedia.org
173. 1961 Goldsboro B-52 Crashthisdayinaviation.com
174. Goldsboro, 1961pbs.org
175. 1966 Palomares accidenten.wikipedia.org
176. 1966 Palomares B-52 crashmilitary-history.fandom.com
177. Palomares,hibakusha-worldwide.org
178. Lost Nuclear Weapon: How the U.S. Misplaced an Atomic Bombtime.com
179. 1983 Soviet nuclear false alarm incidenten.wikipedia.org
180. 1983 Soviet nuclear false alarm incidentmilitary-history.fandom.com
181. The Man Who "Saved the World" Dies at 77armscontrol.org
182. The Man Who Helped Avert Nuclear Armageddonhistory.com
183. Acute Radiation Syndrome - Stanfordlarge.stanford.edu
184. Acute Radiation Syndromes [ Hematopoietic, GI, CNS ] For Radiologic Technologists • How Radiology Workshowradiologyworks.com
185. A Brochure for Physicians: Acute Radiation Syndromecdc.gov
186. Acute Radiation Syndrome - MalaCardsmalacards.org
187. Acute Radiation Syndrome - Large.stanford.edu.large.stanford.edu
188. Rad (radiation unit)en.wikipedia.org
189. Solid Cancer Incidence among the Life Span Study of Atomic Bomb Survivors: 1958–2009bioone.org
190. Summary of radiation effects on incidence of solid cancers in the Life Span Study of atomic bomb survivors: 1958–2009 - PMCpmc.ncbi.nlm.nih.gov
191. Studies of Atomic Bomb Survivors - Amy Berringtongrantome.com
192. Radiation and Risk of Liver, Biliary Tract, and Pancreatic Cancers among Atomic Bomb Survivors in Hiroshima and Nagasaki: 1958–2009 (Journal Article)osti.gov
193. EXECUTIVE SUMMARYrerf.or.jp
194. (PDF) Changes in Atomic Bomb Survivors' Dosimetry with the New Dosimetry System DS02researchgate.net
195. DS02: A New Dosimetry System for A-Bomb Survivor Studiesenergy.gov
196. Solid Cancer Incidence among the Life Span Study of Atomic Bomb Survivors: 1958–2009 - PMCpmc.ncbi.nlm.nih.gov
197. Acute Radiation Syndromemobile.fpnotebook.com
198. Changes in atomic bomb survivors' dosimetry with the new dosimetry system DS02 (Journal Article)osti.gov
199. International humanitarian law and nuclear weapons: Irreconcilable differences - Bulletin of the Atomic Scientiststhebulletin.org
200. The ICJ Advisory Opinion (J) - Nuclear Weapons Lawcambridge.org
201. Advisory Opinion on the Legality of the Threat or Use of Nuclear Weaponsen.wikipedia.org
202. Advisory Opinion of 8 July 1996icj-cij.org
203. International Court of Justice and its 1996 advisory opinionreachingcriticalwill.org
204. The Treaty on the Prohibition of Nuclear Weapons At A Glancearmscontrol.org
205. Treaty on the prohibition of nuclear weaponseuroparl.europa.eu
206. Treaty on the Prohibition of Nuclear Weaponsdisarmament.unoda.org
207. Treaty on the Prohibition of Nuclear Weaponsunfoldzero.org
208. Looking Back: The 1996 Advisory Opinion of the International Court of Justicearmscontrol.org
209. Advisory Opinion on the Legality of the Threat or Use of Nuclear Weapons :: International Case Law, Court Opinions & Decisions :: Justialaw.justia.com
210. States-Parties Meet on Nuclear Arms Ban Treatyarmscontrol.org
211. Treaty on the Prohibition of Nuclear Weapons (TPNW)nti.org
212. How the Treaty on the Prohibition of Nuclear Weapons Impacts the United States, and Why the United States Must Embrace its Entry into Forcejia.sipa.columbia.edu
213. Nuclear weapons and international humanitarian lawicrc.org
214. Ukraine Symposium - Lawful Use of Nuclear Weapons - Lieber Institute West Pointlieber.westpoint.edu
215. The controversy around nuclear weapons — Public International Law & Policy Grouppublicinternationallawandpolicygroup.org
216. U.S. Nuclear Modernization Programsarmscontrol.org
217. Inside the Pentagon’s High-Stakes Nuclear Overhaulthecipherbrief.com
218. UNCLASSIFIED 1 UNCLASSIFIED STATEMENT OF THE HONORABLE ELBRIDGE A. COLBY,media.defense.gov
219. Russian nuclear weapons, 2026 - Bulletin of the Atomic Scientiststhebulletin.org
220. Putin's Nuclear Crisis - May 2026globalsecurity.org
221. Signal and Substance: Russia's Nuclear Modernisation After New STARTgreydynamics.com
222. Chinese nuclear weapons, 2026 - Bulletin of the Atomic Scientiststhebulletin.org
223. Putin hails Russia’s test launch of ‘most powerful missile in the world’aljazeera.com
224. Arms Control and Proliferation Profile: Chinaarmscontrol.org
225. NMHB 2020 [Revised] - Chapter 15acq.osd.mil
226. Uranium enrichment: A chemist explains how the surprisingly common element is processed to power reactors and weaponstheconversation.com
227. armscontrolcenter.org Uranium Enrichment: For Peace or for Weaponsarmscontrolcenter.org
228. Fissile Materials Basicsucs.org
229. Nonproliferation and fusion power plantsarxiv.org
230. Governance of Dual-Use Technologies: Theory and Practiceamacad.org
231. IAEA SAFEGUARDS GLOSSARY 2001 Edition INTERNATIONAL NUCLEAR VERIFICATION SERIESwww-pub.iaea.org
232. Project Plowshareen.wikipedia.org
233. Peaceful nuclear explosionen.wikipedia.org
234. Project Plowshare: The Peaceful Use of Nuclear Explosives in Cold War Americadirect.mit.edu
235. Enriched Uranium - an overviewsciencedirect.com
236. Project Plowshare programosti.gov
237. Project Gasbuggyen.wikipedia.org
238. Operation Plowshare: “Peaceful” Nuclear Explosions - Daily Kosdailykos.com
239. “Peaceful” nuclear explosives? - Physics Todayphysicstoday.aip.org
240. Uranium Enrichment - World Nuclear Associationworld-nuclear.org
241. IAEA Safeguards Serving Nuclear Non-Proliferationiaea.org
242. The Open-Weight Paradox: Why Restricting Access to AI Models May Undermine the Safety It Seeks to Protectarxiv.org
243. Safeguards to Prevent Nuclear Proliferation - World Nuclear Associationworld-nuclear.org