Flávio Bolsonaro remains numerically ahead of President Luiz Inácio Lula da Silva in 12 of the 13 latest national runoff polls compiled by Poder360. But two new surveys released on Monday, October 19, show his lead shrinking with just six days until Brazil’s presidential runoff. That combination points to a competitive finish — not a settled result.
The apparent contradiction is the defining feature of the race. A broad collection of polling institutes puts the right-wing senator in front, yet the margins reported by several of those institutes are small enough to fall within their stated sampling errors.
The more useful question is not simply who leads in the latest headline, but whether changes within the same polling series show a consistent movement.
Why the “12 of 13” figure needs context
In a comparison published October 19, Poder360 examined 13 recent nationwide studies, with interviewing dates spanning October 7–18.
Flávio Bolsonaro had a numerical advantage in 12. Lula led the remaining poll, conducted by Vox Brasil.
That is a count of polls showing a numerical lead, not a weighted polling average, a forecast or a statement that Flávio has a 12-in-13 probability of winning.
The 13 studies differ in sample sizes, interview dates, sampling methods and uncertainty. Counting their leaders is useful as a descriptive snapshot, but it should not be mistaken for a statistical model.
The latest polls show both the lead and the uncertainty
| Pollster | Flávio Bolsonaro | Lula |
|---|---|---|
| PoderData/Aya | 52.0% | 48.0% |
| BTG/Nexus | 50.3% | 49.7% |
| AtlasIntel/Bloomberg | 51.1% | 48.9% |
| Vox Brasil | 49.0% | 51.0% |
Valid-vote percentages reported by Poder360 on October 19. These are separate surveys, not a Votendi average or an election forecast.
BTG/Nexus surveyed 2,002 respondents from October 16–18 and reported a two-percentage-point margin of error. Its 50.3%–49.7% split is therefore a statistical tie within the survey’s stated uncertainty.
AtlasIntel interviewed 4,998 people from October 13–18, reporting a one-point margin of error. Its valid-vote split was 51.1%–48.9%.
PoderData/Aya, meanwhile, showed a wider 52%–48% advantage for Bolsonaro among valid votes.
Where Lula has gained ground
Comparisons within individual polling series suggest that Bolsonaro’s lead has contracted in at least two recent surveys.
In the Nexus results for all respondents’ stated choices, Bolsonaro’s margin declined from 47.9% versus 45.1% in an earlier October survey to 46.5% versus 45.9% in the latest one — from 2.8 to 0.6 percentage points.
AtlasIntel/Bloomberg also recorded a tighter race on its comparable total-vote measure: Bolsonaro’s 51.1% to 45.7% lead in an earlier survey became 49.8% to 47.6% in the latest study.
The gap narrowed from 5.4 to 2.2 points.
These within-series comparisons are more informative about direction than treating different pollsters’ estimates as consecutive points in one trend.
Nevertheless, the changes do not establish that Lula has overtaken his challenger. Different surveys continue to show different margins, and movements between polling waves must be interpreted in light of statistical uncertainty.
What could decide the final six days?
The runoff follows a first round in which Flávio Bolsonaro finished narrowly ahead of the incumbent.
Both candidates now face a different electoral test: persuading voters whose first-round choices are no longer on the ballot and ensuring that supporters actually turn out on October 25.
The first runoff debate between Lula and Bolsonaro took place on October 18. Most newly published polling was conducted before or during that day, so it would be premature to attribute the narrowing gap to the debate.
Subsequent surveys will be better positioned to test whether any change persisted after the confrontation.
Turnout matters as much as small movements in expressed voter preference. A national survey measures intentions among its sampled electorate; the final outcome depends on which voters participate and how preferences evolve before polls close.
How Votendi is tracking the race
Votendi separates the results of individual polling organisations from its own tracker calculations.
The “12 of 13” finding in Poder360 is a descriptive survey comparison and should not be read as Votendi’s polling average.
Our tracker records pollsters, fieldwork dates, samples and individual results so readers can compare how estimates differ and change over time.
With the October 25 runoff approaching, the central question is whether Lula’s recent improvement appears again in surveys using comparable methods — or whether Bolsonaro maintains a narrow but persistent advantage.
Follow the Brazil 2026 election tracker for the latest published surveys. This analysis builds on Votendi’s October 19 update on X.
Sources: Poder360 and Reuters, October 19, 2026. Polling estimates are not official election results or forecasts.
