Opinion polling for the 2026 Brazilian presidential election measures voters’ stated preferences among actual or hypothetical presidential candidates. These opinion polls include first-round voting-intention surveys, simulated second-round contests, and questions about candidate rejection or willingness to consider particular contenders. They describe preferences during their fieldwork period rather than establish the eventual ballot or predict an election result with certainty.
Electoral context
Brazil elects its president through a two-round system. A candidate wins in the first round by obtaining an absolute majority of valid votes; otherwise, the two leading candidates contest a runoff. Blank and invalid votes are excluded from the valid-vote total.
This structure makes first-round and runoff polling distinct exercises. A first-round survey measures preferences across a broader candidate list. A runoff survey presents a particular pair, asking voters to choose between them even when neither was their initial preference. Several hypothetical runoff pairings may therefore appear in the same polling release.
Luiz Inácio Lula da Silva, who began his third presidential term in January 2023, has been included in surveys testing possible 2026 contests. Inclusion in a questionnaire does not itself confirm a candidacy. Candidate lists in early surveys can change as contenders and political parties make decisions about nominations and alliances.
Principal measures
Presidential polling commonly distinguishes the following measures:
- Spontaneous voting intention: respondents name a preferred candidate without being shown or read a list.
- Prompted voting intention: respondents select from a specified list of names.
- Runoff voting intention: respondents choose between two named contenders.
- Candidate rejection: respondents identify candidates for whom they would not vote, according to the question’s wording.
- Government approval: respondents evaluate the incumbent administration rather than choose a presidential candidate.
These measures are not interchangeable. Spontaneous questions are especially sensitive to name recognition, while prompted results depend partly on which candidates are offered. Government approval may be associated with electoral preferences, but an approval percentage is not a candidate’s vote share.
Methodology and comparability
Surveys may use face-to-face interviews, telephone interviews, or online recruitment. Their estimates depend on sampling procedures, geographical coverage, respondent selection, and adjustments intended to make the sample resemble the electorate. Such adjustments may draw on demographic characteristics including age, education, sex, and region.
A meaningful comparison requires more than matching publication dates. Important details include:
| Detail | Importance |
|---|---|
| Fieldwork dates | Identify when preferences were measured, which may precede publication by several days. |
| Candidate list | Different lineups can redistribute support among alternatives. |
| Interview method | Different methods may reach different groups or elicit different responses. |
| Sample size and design | Affect statistical precision and the interpretation of reported uncertainty. |
| Treatment of undecided, blank, and invalid responses | Determines the denominator used to calculate percentages. |
| Questionnaire wording and order | Can affect how respondents interpret and answer questions. |
A national poll is not directly interchangeable with a survey conducted in one state or municipality. Likewise, subgroup estimates generally have greater uncertainty than estimates for the full sample.
Reporting and interpretation
Brazilian electoral legislation requires election-related polls intended for public dissemination to be registered with the electoral authorities from 1 January of the election year, at least five days before publication. Registration information includes details about commissioning, financing, methodology, and the questionnaire. Registration is a disclosure requirement, not official certification that a survey is accurate.
Reported margins of error describe only part of measurement uncertainty. Coverage gaps, nonresponse, weighting choices, and question effects can also influence results. The statement that candidates are “technically tied,” often encountered in news reporting, does not mean that their underlying support is necessarily equal.
Percentages calculated from all respondents must also be distinguished from estimates restricted to valid voting intentions. Excluding undecided, blank, and invalid responses changes the denominator and increases the remaining candidates’ reported shares.
Early hypothetical scenarios are particularly sensitive to changes in the candidate field. Polls using different lineups should therefore not be combined into a single trend without accounting for those differences. Even comparable surveys measure stated preferences, not turnout, final voter decisions, or the certified vote count.