A new Gerp poll shows Senator Flávio Bolsonaro leading President Luiz Inácio Lula da Silva in a hypothetical second-round presidential runoff [1].
The results highlight a deepening divide in Brazilian voter sentiment as the country weighs the potential for a return to the political lineage of the former president.
According to the survey, Flávio Bolsonaro holds 45% of voting intentions [1], while Lula follows with 42% [1]. The poll also found that nine percent of respondents would not vote for either candidate [2], and four percent did not know how to answer [2].
The Gerp survey was fielded from July 3 to July 7, 2026 [1, 2]. It utilized a sample size of 2,000 people [1], with a reported margin of error of ±2.2 percentage points [1].
These findings contrast with other available data. A separate poll reported by MSN indicates a different outcome, placing Lula in the lead with 45% of voting intentions compared to 40% for Flávio Bolsonaro [7, 8].
Because the results between different polling firms vary, the current standing of the candidates remains contested. The gap in the Gerp poll is narrow enough to be influenced by the margin of error, creating a statistical tie in some interpretations.
The disparity between the two polls suggests that voter loyalty remains fluid. The three percent difference in the Gerp poll [1] and the five percent lead for Lula in the alternative survey [7, 8] reflect a highly polarized electorate where small shifts in undecided voters could change the projected winner.
“Flávio Bolsonaro holds 45% of voting intentions”
The conflicting data between the Gerp and MSN polls indicates a lack of consensus on the current political trajectory in Brazil. With margins of error overlapping and results swinging between the two primary figures, the race remains highly volatile. This volatility suggests that neither candidate has secured a dominant lead, and the outcome of a potential runoff would likely depend on the ability of either camp to capture the small percentage of undecided or resistant voters.
