Brazil Runoff Today: A 1-Point Race Decides It

Brazilians are voting today, Sunday, October 25, in a presidential runoff that every major final poll places inside its own margin of error. I have covered a lot of close elections for this site, and this is the rare one where the honest headline is that nobody knows. The first round on October 4 put Senator Flavio Bolsonaro first and President Luiz Inacio Lula da Silva second, close enough that the runoff was always going to turn on who shows up, not on who persuades. The tracking polls since then, summarized in Ipespe/Abrapel runoff tracking reported October 23, show the gap compressing week by week. Election-day reporting in today's runoff coverage puts close to 160 million people eligible to vote, with voting mandatory.

The first round left almost nothing to split

The official first-round shares were 47.03 percent of valid votes for Bolsonaro and 45.16 percent for Lula, a gap of 1.87 points. In raw votes that is about 56.1 million to 53.9 million, with every other candidate together taking roughly 9.3 million. Neither man cleared the majority needed to win outright, so the second round is a two-person race for a pool that is mostly already decided, plus the voters who stayed home the first time. I keep coming back to that 9.3 million figure from our earlier Brazil first-round runoff math: it is larger than the gap between the finalists by a factor of four, and most of it will not move as a bloc.

Figure 1: First-round votes in millions. Source: TSE results as reported by O Globo and MercoPress.

The tracking poll narrowed, it did not flip

Within the single Ipespe series, Bolsonaro led by 5.4 points in the October 6 to 8 round, by 2.4 points in the October 13 to 15 round, and by 1 point, 50.5 to 49.5, in the October 20 to 22 round. Read that carefully. A shrinking gap inside one pollster's series is evidence of compression, not proof of a lead change, and the final surveys from other houses landed in the same place: effectively tied, with a nominal edge of about a point in several of them. That is the same discipline I applied to the IMF's Middle East forecast revisions last week: a trend inside a series is real, the level still carries uncertainty.

Figure 2: Ipespe/Abrapel runoff tracking, share of valid votes. Source: Ipespe/Abrapel, October 2026.

Poll (final rounds)	Bolsonaro	Lula	Gap
First round, official (valid votes)	47.03%	45.16%	1.87 pts
Ipespe, Oct 6-8	52.7%	47.3%	5.4 pts
Ipespe, Oct 13-15	51.2%	48.8%	2.4 pts
Ipespe, Oct 20-22	50.5%	49.5%	1.0 pt

Turnout is the variable that decides this

About 33 million eligible voters did not vote in the first round, on reporting that puts abstention at roughly a fifth of the electorate. In a race separated by 1 point, moving even a small share of those non-voters matters more than any last debate moment. No candidate who finished second in the first round has gone on to win a Brazilian presidential runoff in the democratic era, which is a real historical pattern, not a law, and today is exactly the kind of near-tie where such patterns break. Polls close at 5 p.m. Brasilia time and Brazil's electronic count is expected to produce a result within hours, so by tonight this analysis will be testable against the actual count.

Figure 3: Brazil (orange) with Brasilia marked. Source: Natural Earth.

What I am watching tonight

First, the abstention rate against the first round's roughly 21 percent. Second, whether the winner's valid-vote share lands inside the final polls' tied range or outside it, because that tells us whether the compression story was right. Third, the margin itself: a winner under 1 point governs a country of more than 200 million with the weakest mandate the runoff system can produce, and that shapes everything from cabinet bargaining to the street politics we tracked in the Qatar LNG buying spree style of energy diplomacy Brazil will now conduct. I will update this page when the electoral court calls it.