Brazil’s Final Polls Show Lula 45%, Flávio 42%
Brazil · Election 2026
Key Facts
—Who. Brazilian voters pick a president on Sunday, October 4. Lula (PT) and Flávio Bolsonaro (PL) lead the final polls.
—What. Datafolha has Lula on 45% of valid votes and Flávio on 42%. Quaest has 46% to 45%. Both polls were published on Saturday, October 3.
—Runoff. Datafolha has Lula 47%, Flávio 46%. Quaest has Flávio 44%, Lula 42%. The margin of error is 2 points in both.
—Why it matters. A candidate needs more than 50% of valid votes to avoid a runoff on October 25. No poll puts anyone above that line, so the next three weeks matter.
The last polls before Brazil votes show a statistical tie. Datafolha has President Lula on 45% of valid votes and Senator Flávio Bolsonaro on 42%; Quaest has 46% to 45%. For readers in the United States, polls open at 7 a.m. Eastern on Sunday and the count starts at 4 p.m. Eastern.
Two Saturday Polls, Two Different Runoffs
Datafolha published its final survey on Saturday evening, October 3. It interviewed 4,006 voters in person, with a margin of error of 2 percentage points (TSE registration BR-01708/2026). Among all voters, Lula has 42% and Flávio Bolsonaro 40%. Once blank and null ballots are removed, the “valid votes” the electoral court counts, it is 45% to 42%.
Quaest, commissioned by Globo, interviewed 3,702 voters on October 2 and 3, also with a 2-point margin (BR-02197/2026). It has Lula on 40% and Flávio on 38% of all voters, or 46% to 45% of valid votes.

The runoff simulations point in opposite directions. Datafolha has Lula ahead 47% to 46%. Quaest has Flávio ahead 44% to 42%. Both gaps sit well inside the margin of error, so neither poll can call a leader.
How the Final Week Moved
Compared with their previous waves, both pollsters show Flávio Bolsonaro gaining. Datafolha’s first-round figure for him rose from 38% (published October 1) to 40%, while Lula held at 42%. Quaest’s rose from 34% (published September 28) to 38%, while Lula moved from 39% to 40%.
In the runoff, Datafolha’s Lula lead narrowed from 48% to 45% (published October 1) to 47% to 46%. Quaest has Flávio ahead by two points. The shift is small, but it all runs in one direction.
Other houses published between September 28 and October 3 span the same band. CNT/MDA (2,007 voters, margin 2.2 points) has Lula ahead 47.8% to 42.1% of valid votes. PoderData/Aya (4,000 voters, margin 1.5 points) has 45% to 43%. AtlasIntel (5,005 voters, margin 1 point) has Lula 45.3% to 42.2%. Our poll tracker lists every survey with its sample and field dates.
Why a Few Points Mean Less Than They Seem
A margin of error of 2 points applies to each candidate’s figure. A 2-point gap among all voters is therefore inside it. Pollsters also reach voters in different ways: in person, by phone, online or with automated calls. That is why the same race produces readings as far apart as Lula +5.7 (CNT/MDA) and Lula +1 (Quaest), both in valid votes.
Polls have missed in Brazil before. Nobody can say which way any error will run this time, or whether there will be one.
What Happens After the Count
Any candidate with more than 50% of valid votes wins on Sunday. Every Brazilian presidential election since 2002 has gone to a runoff. If that happens again, the top two meet on Sunday, October 25. The official count from the electoral court (TSE) starts at 4 p.m. Eastern; follow it on our live results page. The complete election guide covers the candidates and the rules.
The Smaller Candidates Could Decide the Runoff
Behind the two leaders, the field is thin. In Datafolha’s valid-vote count, Ronaldo Caiado (PSD) has 4%, Augusto Cury (Avante) 3%, Renan Santos (Missão) 3% and Romeu Zema (Novo) 1%. Quaest has Cury, Santos and Caiado on 3% each. Together they hold roughly a tenth of the valid vote.
In a runoff that tenth matters. In Datafolha’s runoff question, 2% of respondents were undecided and 6% said they would vote blank, null or for neither candidate. Both campaigns will court those voters, along with the supporters of the eliminated candidates, between October 5 and 25.
What It Means for You
If you hold Brazilian assets or travel to Brazil, the poll picture says to expect a close race and possibly three more weeks of campaigning. Brazilian markets reopen on Monday, October 5. Our piece on what the vote means for US investors walks through the real, the Ibovespa and the US-listed Brazilian stocks.
What Is Not Known
Polls do not predict turnout, and they cannot say how many voters who say they are undecided will switch. A poll is a snapshot of opinion, not a result. The official figures come only from the TSE.
What do the final Brazil polls show?
Datafolha, published on October 3, has Lula on 45% of valid votes and Flávio Bolsonaro on 42%. Quaest, also published on October 3, has 46% to 45%. Both are statistical ties within a 2-point margin of error.
Who leads the runoff polls?
The two final polls disagree. Datafolha has Lula ahead 47% to 46%, while Quaest has Flávio Bolsonaro ahead 44% to 42%. Both gaps are inside the margin of error.
What are valid votes?
Valid votes are the votes cast for candidates, with blank and null ballots removed. The electoral court counts the 50% needed to win outright on this basis.
Can a candidate win on Sunday without a runoff?
Yes, with more than 50% of valid votes. No final poll puts either leader at that level. If nobody clears it, the top two meet on Sunday, October 25.
When will the results come in?
Polls close at 5 p.m. Brasília time, which is 4 p.m. Eastern. The electoral court starts publishing the count from that moment, and Brazil counts electronically, so a national picture usually emerges within hours.
Sources
Datafolha (poll BR-01708/2026, 3 October) and Quaest/Globo (BR-02197/2026, 3 October), as reported by Folha, O Povo, DGABC and InfoMoney; CNT/MDA, PoderData/Aya and AtlasIntel as listed in our poll tracker; Tribunal Superior Eleitoral (TSE).
Frequently Asked Questions
This article was drafted with automated assistance and reviewed before publication. How we use AI · Report an error · Editorial responsibility: Matthias Camenzind, Editor-in-Chief