On Sunday, October 25, Brazil encapsulated in a single election the direction in which Latin American democracies are heading. Lula da Silva won the runoff with 51.76% of the vote, against Flávio Bolsonaro’s 48.24%, a margin of 3.52 percentage points. This was the first time since the return of direct elections that a candidate had come from behind to win a runoff in Brazil. In the seven previous runoffs since 1989, the candidate who led the first round had always won. On this occasion, turnout in the first round had already reached a record high. Only 1.87 percentage points separated the two leading candidates, who together secured 92.2% of the vote. Hours after the runoff, Flávio Bolsonaro said he did not recognize the result and announced that he would seek to have the election annulled, without providing any evidence. In the end, he did not do so.
The runoff system was designed to ensure that presidents took office with a majority and legitimacy. Today, it does three other things: it has ceased to be an exception and become the rule, manufactures majorities and rewards rejection, and shifts the contest to the day after the election.
The runoff has become the norm
In countries with a two-round system, the runoff has ceased to be exceptional. In the 1990s, 52% of elections were decided in a second round; in the 2000s, 54%; in the 2010s, 71%; and so far in this decade, 80% (16 out of 20). In several countries, it has already become the norm: Guatemala has held a runoff in every election since 1995, Peru in every election since 2000, and Chile since 1999. Brazil has held seven consecutive runoffs since 2002, Colombia five since 2010, and Ecuador four since 2017.
The driving force is fragmentation. In elections that went to a runoff, the two most-voted candidates together secured 71.3% of the vote in the first round during the 1990s, compared with just 57.1% in the 2020s. In those elections, the candidate who wins the first round now receives a median of just 32.7% of the vote. Behind this fragmentation lies a crisis of representation. According to the Latinobarómetro 2026, only 15% of citizens trust political parties, and 76% do not feel represented by their Congress. When parties fail to aggregate preferences, the runoff ends up doing the work they have stopped doing: building a majority where none exists.
Manufactured majorities, elections of rejection
The runoff produces a presidential majority, but not a legislative one. Those who have won a runoff in this decade received an average of 25.8% of the vote in the first round: half took office with the initial backing of one in four voters or fewer. The most extreme cases are Alejandro Giammattei, who advanced to the runoff with 13.96%; Bernardo Arévalo, with 15.51%; Rodrigo Chaves, with 16.78%; and Keiko Fujimori, with 17.19%.
Moreover, the second round no longer simply confirms the first. Come-from-behind victories have become more frequent: after just two out of 13 runoffs in the 2000s, there were six out of 20 in the 2010s and, in the 2020s, including Brazil, half of all runoffs (eight out of 16). The narrower and more fragmented the first-round result, the more likely a comeback becomes. With fragmented first rounds and polarized electorates, many people vote in the second round for the candidate they reject the least, rather than the one they prefer.
This has two consequences. First, the winner’s legitimacy is, in part, borrowed: a substantial share of their votes does not come from their own support base. Second, the runoff penalizes incumbents: governing parties win 58% of elections decided in the first round, but only 30% of those that go to a second round. In this decade, that penalty has intensified, and the old argument that the runoff moderates electoral competition no longer withstands the evidence either: Bolsonaro, Castillo, Milei, Kast, and De la Espriella have won polarized runoffs.
The contest shifts to the day after
Runoff elections are becoming increasingly close. In this decade, before Brazil, the average margin between the two finalists was 4.8 percentage points, compared with 8.1 across the entire period studied, and the region’s seven closest runoffs have all taken place since 2014. With margins this narrow, any incident can become an argument for refusing to recognize the result. This is where a paradox emerges. According to the Latinobarómetro 2026, 74% of citizens believe there can be no democracy without elections, but only 34% trust their electoral authorities. We believe in elections, but not in those who organize them.
Brazil offers an important lesson. The conflict did not originate with the electoral referee: a Supreme Federal Court judge extended voting hours by two hours, but only at polling stations that were still open, by which time thousands of ballot boxes had already closed. The president of the Superior Electoral Court (TSE) acknowledged that such an extension had never occurred before. The TSE president, Kassio Nunes Marques, maintained that thousands of ballot boxes had already closed when the decision was made, and the court temporarily suspended the receipt of results until after 7 p.m. That decision handed the losing candidate the argument he had been looking for. A good referee is not enough: electoral integrity depends on the entire ecosystem, including other actors respecting the boundaries of their respective powers.
Where are we headed?
Citizens retain a reservoir of democratic commitment, but it has cracks. According to Latinobarómetro (2026), support for democracy stands at 52%, but only 30% are satisfied with how it works. In fact, 63% accept “a bit of a strong hand.” With 41% support for democracy, Brazil has the second-lowest level in the region, ahead of only Guatemala.
The proposed solution does not necessarily involve eliminating the runoff, which effectively addresses the problem of legitimacy at the outset of a president’s term. Rather, it involves safeguarding the system surrounding it: the ecosystem of electoral integrity, as emphasized in recent International IDEA reports. Hence the following proposals: first, reform the institutional framework surrounding the runoff and hold legislative elections concurrently with the second round, so that the president does not take office without support in Congress; second, shield electoral bodies from interference and respect their exclusive authority to administer the electoral process, including in their dealings with the courts; and third, require political parties to commit, publicly and in writing, before voting takes place, to recognizing the results.
The runoff was designed to elect presidents with a majority. Today, it faces a different test: ensuring that losing by three percentage points is also acceptable. In excessively competitive and polarized contexts, recognition of election results becomes the central issue. And that is not determined by the referee alone, but by everyone who takes part in the game.
Autor
Researcher at the Institute of Legal Research of the National Autonomous University of Mexico. PhD from the University of Salamanca and Master's in Latin American Studies from the same institution. Founder and coordinator of the Network of Women Political Scientists - #NoSinMujeres (Not Without Women)
