The presidential election, analysed

The count after 17:00

Polls closed at 17:00, Brasília time. The times below, in Brasília time, are those of the site's readings of the TSE's totals through the night, not of each section's release. The runoff's first votes appeared at 17:07; half of the sections were counted by 20:28, 90% by 21:03, and the last one at 23:23.

Lula led first, Flávio Bolsonaro went ahead at 19:45, and Lula took the lead back at 20:25, never to lose it: the lead changed hands 2 times.

The race was called at 21:28. After that, 158,328 more valid votes were counted: 108,970 for Lula and 49,358 for Flávio Bolsonaro.

The busiest half hour was from 20:30 to 21:00: 42 million valid votes, 53.39% of them for Lula.

In the 1st round, the count began at 17:22 and ended at 03:00.

Who decided the election

Lula won by 4,430,191 votes. The 2,875 municipalities he carried gave him 18 million votes more than Flávio Bolsonaro; Flávio Bolsonaro's 2,695 took back 13 million.

Lula's 43 biggest net municipalities alone add up to more than the final margin. On the other side, it would take Flávio's 71 biggest to match it.

The state capitals gave Lula a net 1,562,067 votes; the other municipalities, 2,839,467; voters abroad, 28,657.

State by state, against Brazil

From the 1st round to the runoff, the gap between the two moved 5.45 points toward Lula in Brazil. It moved further than in the country in Rio de Janeiro (+1.95), Alagoas (+1.81), Sergipe (+1.49); less in Rondônia (−3.11), Acre (−2.76), Santa Catarina (−2.68).

Against the 2022 runoff, Brazil moved 1.78 points toward the PT. It went furthest in Federal District (+7.28), Amapá (+6.70), Acre (+5.68). Against 2022, each state's difference from Brazil is also how far it missed a "uniform swing": the forecast that every place would move as the country did.

Municipality by municipality: before and after

Each dot is a municipality: across, the earlier vote (the PT in the 2022 runoff, or Lula between the two finalists in the 1st round); up, Lula in the 2026 runoff, between the two. On the diagonal, nothing changed; on the line parallel to it, the municipality moved exactly as Brazil did.

Against 2022, 692 municipalities ended above Brazil's line (they moved more toward Lula than the country) and 4,878 below. The correlation between the two votes, weighted by electorate, is 0.985: the 2026 map is largely 2022's.

Against the 1st round, 2,083 ended above and 3,488 below (correlation 0.999).

The municipalities that turned

From the 1st round to the runoff, 212 municipalities changed winner between the two finalists: 211 turned to Lula (11 million voters) and 1 to Flávio Bolsonaro (3,080 voters).

Against the 2022 runoff, 31 municipalities went from the PL to the PT (7.2 million voters) and 278 from the PT to the PL (2.2 million voters); 5,258 kept their side.

By municipality size, capitals and regions

Lula took 54.22% in municipalities of up to 10,000 inhabitants and 53.25% in those over 1 million. In the state capitals, 52.75%; in the other municipalities, 51.49%.

From the 1st round to the runoff, the gap moved 3.88 points toward Lula in the smallest municipalities and 6.14 points in the largest. Each group is its municipalities' votes added up, not an average of them. Population: IBGE (5 bands).

The map by the size of the electorate

On an ordinary map, each municipality takes up the area of its territory, and the vast municipalities of the North and Centre-West seem to weigh more than the big cities. Here each municipality is a circle with an area proportional to its electorate, near where it is on the map.

Lula won in 2,875 municipalities, with 49.99% of the electorate; Flávio Bolsonaro, in 2,695, with 50.01%.

Polarization: 2022 and 2026

In 2026, 52.6% of the electorate lived in municipalities decided by 20 points or more (60% to 40%), against 52.8% in 2022; in municipalities won 70% to 30% or more, 21.3% (2022: 22.0%).

The average distance from a tie, weighted by electorate, was 24.53 points in 2026 and 25.27 points in 2022. Both runoffs set the same parties against each other, PT and PL: the comparison is between elections of the same kind, in the 5,570 municipalities that voted in both.

Where the 1st round's votes went

Estimate: of Escritor Augusto Cury's 3.4 million voters, 52% would have voted for Lula, 30% for Flávio, 15% blank or null and 4% didn't vote; of Renan Santos's 2.7 million voters, 68% would have voted for Lula, 25% for Flávio, 5% blank or null and 1% didn't vote; of Ronaldo Caiado's 2.6 million voters, 27% would have voted for Lula, 55% for Flávio, 13% blank or null and 6% didn't vote.

Of those who voted blank or null in the 1st round, 61% would have chosen Lula in the runoff and 0% Flávio. Of those who didn't vote, 86% didn't vote in the runoff either; 8% turned out for Lula and 4% for Flávio.

This is a statistical estimate, not a count: the vote is secret, and the TSE publishes only each place's totals. The ecological regression (Goodman) assumes that, within each state, a candidate's voters split the same way in every municipality; in brackets, the 90% interval from 200 resamples of the municipalities. The fit explains 88% to 100% of the variation between municipalities.

Abstention and blank and null votes

Abstention was 19.22% in the runoff (30,509,005 voters), against 21.08% in the 1st round and 20.58% in the 2022 runoff. Blank and null votes were 3.24% of the votes cast, against 4.77% in the 1st round and 4.59% in 2022.

Among municipalities with 10,000 voters or more, abstention rose most in Marechal Thaumaturgo (AC), from 21.39% to 40.65%; blank and null votes, in Lagoa da Confusão (TO), from 1.92% to 4.07%.

Abstention is measured as the TSE measures it: over turnout plus abstention. Blank and null votes, over the votes cast.

President and governor on one ballot

In 6 states voters chose a president and a governor in the same runoff. In each municipality, the two votes move together where the correlation is close to 1 (or to −1, when the governor candidate compared is on the other side) and are independent near 0.

Acre (estimate): 75% of Lula's voters and 57% of Flávio's would have voted for Mailza Assis (PP), who was elected.

Amazonas (estimate): 96% of Lula's voters and 2% of Flávio's would have voted for Omar Aziz (PSD), who was elected.

Espírito Santo (estimate): 14% of Lula's voters and 78% of Flávio's would have voted for Lorenzo Pazolini (Republicanos), who was elected.

Rio Grande do Norte (estimate): 71% of Lula's voters and 0% of Flávio's would have voted for Cadu de Lula (PT), who was elected.

Tocantins (estimate): 49% of Lula's voters and 65% of Flávio's would have voted for Vicentinho Júnior (PSDB), who was elected.

Federal District is a single municipality: there's no way to estimate the split ticket from its totals.

Inside the cities: the electoral zones

190 municipalities have more than one electoral zone, and in the biggest cities they voted very differently.

Among the state capitals, the widest gap was in Rio de Janeiro: 60.44 points between zone 16 (Laranjeiras), where Lula had the biggest lead, and zone 9 (Recreio).

How each number was computed

Source: the Tribunal Superior Eleitoral (TSE) totals by municipality, electoral zone and city abroad, read once each round's last section was counted. 2022's are that election's official results.

Shares to two decimals, rounded as the TSE rounds them (half up). Every change is the difference of the figures shown: the 3.58-point lead is 51.79 − 48.21.

Change in the gap: how much the distance between the two finalists moved, in points of the valid vote. From the 1st round, against each one's vote among every candidate; from 2022, against the PT–PL runoff. Against Brazil: the place's change minus the country's.

Turned: the place changed winner between the two finalists (in the 1st round, whichever of the two was ahead of the other, whoever led overall); a tie isn't a turn.

Vote transfer and split ticket: Goodman's ecological regression, weighted by electorate, with each row held to a valid split (each part between 0% and 100%, adding up to 100%), fitted state by state and added up; 90% interval by bootstrapping the municipalities. It is an estimate: it assumes that, within a state, each group split the same way in every municipality.

Cartogram: Dorling circles, with an area proportional to the runoff's electorate, pushed apart until none overlap. Population strata: IBGE's estimate for each municipality.

Boa Esperança do Norte (MT), installed in 2025, has no 1st-round file of its own: its figures are its state's total minus the other municipalities. It didn't exist in 2022. A few municipalities' electorates change by a few voters between the rounds (transfers and voting away from home). More at /en/sobre.

Download the data

Every number on this page comes from these files, in CSV (comma-separated, UTF-8, counts only: every share can be recomputed). In Excel, open them through Data › Get Data › From Text/CSV. More formats and the API at https://eleicoes2026brasil.com/en/dados.

• President by municipalityEach row: a municipality: 1st round, runoff and the 2022 electionpresidente-municipios.csv
• President by stateEach row: a state or abroad: 1st round, runoff and 2022presidente-estados.csv
• President by electoral zoneEach row: an electoral zone of the municipalities with more than one: 1st round and runoffpresidente-zonas.csv
• President abroadEach row: a city abroad: 1st round and runoffpresidente-exterior.csv
• The presidential count, minute by minuteEach row: the national totals the TSE published at a moment of the night, from 17:00 Brasília time, in both roundspresidente-apuracao.csv
• Governor runoffsEach row: a municipality of the six states with a governor runoffgovernador-2t-municipios.csv