Brazil Election 2026: Runoff Date, Polls, Odds and Results

Brazil's presidential election goes to a runoff between Luiz Inácio Lula da Silva and Flávio Bolsonaro on Sunday, 25 October 2026. As of this guide's check on 19 October, the second-round result has not been decided. The TSE results portal is the place to check the official count when voting ends.

The number beside a candidate's name needs a label. A poll measures reported voting intention among a sample. A market price describes a contract on who wins. The official count records votes already cast. This guide connects those three views, with the Brazil election room for current available market data.

In its 5 October report, Brazil's Superior Electoral Court recorded Flávio Bolsonaro on 47.03% of valid votes and Lula on 45.16%, with 99.99% of polling sections counted at the stated cutoff. Those are first-round shares. Neither reached a majority, so the two advanced.

Finishing first on 4 October did not award the presidency. The runoff is a new vote between the remaining two candidates. Adding the first-round shares of eliminated candidates to one finalist assumes that their voters all move together and that turnout stays the same. Neither assumption is established by the first-round totals.

Download chart figures:CSVJSON with sourcesSVG chart

Use each source for the question it can answer. A polling table is useful for comparing voting intention over time. A market screen is useful for seeing how a priced expectation changes. The electoral authority is the source for the result.

Download table figures:CSVJSON with sources

Begin with the pollster's own release and its fieldwork dates. Our polls-versus-odds explainer uses a named Datafolha survey to show why an all-response percentage changes when the denominator becomes valid votes. It also explains why neither figure is a win probability.

Compare national runoff questions with national runoff questions. A state poll, a first-round poll and a hypothetical matchup from months ago answer different questions. Check whether undecided respondents and blank or null responses are included. A release date alone does not tell you when respondents were interviewed.

The TSE's polling resources link to PesqEle registrations. Look for the registration number, commissioning party and methodology when checking a circulated image. A screenshot stripped of that context is a poor basis for comparing surveys.

Read the contract question before reading the price. A market on the first-round leader, a market on the eventual president and a market on the victory margin have different conditions. A winning first-round position can coexist with an uncertain presidential race.

Then inspect the price's timestamp, the spread between bids and asks, and the depth available near that quote. A headline price does not promise that an order of any size could execute there. Total traded volume is a separate measure from the offers currently available.

The Brazil room connects the election's market boards with market analytics. For the wider mechanics, see our order-size and depth study.

The TSE says result publication starts after voting closes at 17:00 Brasília time. Our results-time guide converts that start into UTC, London, Tallinn and New York time and explains how to follow a partial count. That start time is not a promise of a completed result or a market payout.

New fieldwork can change the polling comparison. Turnout can change the relationship between survey intention and the votes cast. A fresh market move can change the price without showing which new fact caused it. On election night, official returns answer a different question again.

Keep a simple record when comparing updates: what changed, when it was observed, which source reported it, and which measure it belongs to. That makes the next update easier to interpret than a stream of unlabelled percentages.