Will Lula win Brazil's 2026 presidential election?

Lula, seeking a fourth term, finished second in a first round for the first time since 1998 and has to win over voters who chose someone else. His government's most concrete move this week was legislative: the Senate approved the end of the six-day work week on October 14, a measure Lula had pushed to get voted before the runoff. He needs voters to credit him for it, and the campaign is built around that and debt relief rather than his own approval, which Quaest put at 43 percent in September.

Flavio Bolsonaro's task is to protect a lead, not build one. He collected the endorsements of Cury, Caiado and Zema in the first week after the vote, which is the cheap part. The harder part is avoiding a fight on ground that favours Lula, and the Senate vote shows how the right is handling it: the measure passed 75 votes to one, with the Liberal Party's Senate leader voting in favour, so Flavio's camp denied Lula a contrast on the issue rather than opposing it.

Reference class: Brazilian presidential runoffs since direct elections returned in 1989. In all seven (1989, 2002, 2006, 2010, 2014, 2018 and 2022) the candidate who led the first round won the runoff. Lula benefited from that rule in 2022, when he led the first round narrowly and held on. It now runs against him.

The closest analogue remains 2022 with the roles reversed: two near-equal camps and a leader below half the vote. It breaks in two ways. The 2026 first-round gap, under two points by our calculation, is the smallest any Brazilian runoff has started from, so the unbroken streak is a weaker guide than it looks. And in 2022 the final runoff polls overstated the left: Quaest's last 2022 survey had Lula 52 to 48, and he won by less, 50.9 to 49.1. If the same right-leaning miss repeats, a two-point Flavio lead in the polls is worth more than two points on election day.

Version 2 landed at 0.12 with Flavio ahead by about 5 points on valid votes across the first post-round polls. Since then four post-round polls are on the record, our conversion to valid votes where needed: Datafolha 52 to 48 (October 8), AtlasIntel about 53 to 47 (fieldwork October 3 to 8), PoderData 51 to 49 (released October 13, down from 53 to 47) and Quaest 46 to 44 on total votes, about 51 to 49 on valid votes (released October 14). The simple average lead is about 3.5 points; the two newest average about 2.1, our calculation.

We weight each poll by recency and by sample size. AtlasIntel has the largest sample of the four and Datafolha the longest track record, and both show Flavio ahead by 4 points or more; PoderData and Quaest are newer and show about 2. Blending them, we put Flavio's current valid-vote lead at about 3 points. We then add 1.5 points for the right-leaning polling miss seen in the 2022 runoff and the 2026 first round, which pollsters may have partly corrected, for an effective deficit of about 4.5 points. With a runoff polling error of about 4 points on the margin, a normal approximation gives Lula about 0.13, our calculation. The outside view, Laplace smoothing of 0 of 7 runoffs won by the first-round runner-up, gives about 0.11. The two newest polls both moved toward Lula, so we land at 0.15, our calculation, a 3-point rise from 0.12. That is below Kalshi's 0.19 because the first-round order has never been reversed and the larger samples still show a wider gap.