Brazil’s judiciary played a pivotal role in containing Jair Bolsonaro’s authoritarian drive, culminating in his September 2025 conviction for attempting a coup. Unlike in most countries—and in Brazil’s own history—where leaders who undermine democracy typically evade post-tenure punishment, high courts constrained Bolsonaro both during and after his presidency through strategic choices. Equally important, their rulings enabled rivals to weaken Bolsonaro’s grip on power while avoiding open confrontation with powerful congressional forces. This experience offers concrete lessons for democracies confronting similar threats, including those now investigating former presidents who sought to subvert democratic rule.
As traditional coups have become rarer, incumbent-led subversion is now the most common form of autocratization worldwide. Reflecting this shift, democratically elected leaders have attempted more self-coups than ever before — with seventeen episodes, both successful and unsuccessful, recorded since 2016.1 Cases of successful self-coups include Tunisian president Kais Saied’s suspension of parliament in 2021 and subsequent approval, by referendum the following year, of a new constitution that vastly expanded presidential powers, and Salvadoran president Nayib Bukele’s dismantling of judicial oversight and use of emergency powers following his party’s landslide parliamentary victory, also in 2021. Failed self-coups include former Indonesian president Abdurrahman Wahid’s 2001 bid to dissolve parliament and impose martial law and former Ecuadoran president Lucio Gutiérrez’s 2005 attempt to dissolve the Supreme Court and rule by emergency decree.
Across these episodes runs a common thread that reflects a broader pattern: Leaders of self-coups have mostly evaded the imposition of effective judicial sanctions. Successful self-coups produce new autocrats who are essentially immune while in office; and unsuccessful ones rarely result in a former leader’s trial and conviction.
Brazil offers an important counterexample. While in office, President Jair Bolsonaro (2019–23) spent years sowing doubt about Brazil’s voting system and attacking its courts. When he lost at the polls in 2022, he pressured the military top command to prevent the transfer of power. In response, on 11 September 2025, the Federal Supreme Court convicted Bolsonaro and seven high-ranking allies for attempting a coup, which entailed plans to declare a state of emergency and install a militarized electoral body to hold new elections, and for inciting an insurrection calling for military intervention.
How did Brazil’s courts check Bolsonaro when other democracies have let would-be autocrats go unpunished? We argue that the conviction of the former president was the culmination of a series of measures taken by the country’s two high courts — the Federal Supreme Court (or STF, Brazil’s supreme court) and the Superior Electoral Court (or TSE, the high court that oversees national elections) — both during and after Bolsonaro’s presidency, that paved the way for his recent sentencing to more than 27 years in prison. Acting largely in self-defense amid high uncertainty while Bolsonaro was still president, the high courts worked both to limit executive aggrandizement and to prevent his attempts to manipulate the electoral rules, all while empowering his rivals and avoiding direct confrontation with congressional elites. After Bolsonaro left office, the high courts swiftly declared him ineligible to run until 2030 and oversaw several criminal investigations that eventually led to his indictment, trial, conviction, and arrest.
Here, we provide a bird’s eye view of the measures taken by Brazil’s high courts that made Bolsonaro’s sentencing possible. While acknowledging that these courts were not the only forces behind the former president’s downfall and conviction, the courts played a pivotal role in bringing about both outcomes. Put differently: Without their decisions, Bolsonaro would likely have remained in power and beyond the reach of justice.
The responses of Brazil’s courts and the dilemmas they faced thus offer lessons for democracies under threat. The courts’ strategic blend of resolve and caution illustrates how judiciaries might limit elected leaders who turn against democracy in practice. Yet it also shows that courts cannot do it all — they face trade-offs and must avoid conflict with powerful congress members who could jeopardize efforts to contain a populist leader. Nor can the courts act alone — their actions must work in concert with those of civil society, independent and opposition parties and lawmakers, and the media to make legal accountability possible.2
Judicial Responses to (Self-)Coups
Bolsonaro’s sentencing is largely exceptional across various measures. One is Brazil’s own history of coups and self-coups, which have frequently been driven by the military. From the 1889 overthrow of the monarchy to the 1964 coup against a democratic government, there have been more than a dozen such episodes. Some succeeded, such as Getúlio Vargas’s self-coup in 1937. Others failed, including the attempts against democratically elected president Juscelino Kubitschek in the 1950s.3 In all these episodes, the legal outcome was the same: impunity. Many perpetrators flourished under the new regime and were later spared when it collapsed; others received presidential pardons or benefited from congressional amnesties issued soon after the coup attempt. Bolsonaro’s conviction thus marks the first time in Brazilian history that a coup attempt, successful or not, has been punished. That the STF also sentenced three retired army generals and a navy admiral to at least nineteen years in prison reinforces that the September 2025 ruling was a rare break with the country’s historical pattern.
Another way to place Bolsonaro’s conviction in context is by looking at the comparative record. Unlike most recent executive-led power grabs, which unfold through the gradual capture of institutions by hollowing out checks and balances from within — as in Hungary under Viktor Orbán — self-coups are marked by more abrupt ruptures. In such moments, leaders try to cast aside constitutional limits altogether, ruling by decree or invoking emergency powers, for example, while sidelining legislatures or courts. One might therefore expect more blatant violations of the constitutional order to be easier to punish. Yet comparative evidence shows that even these more visible assaults on democracy rarely result in convictions (see Table).
In previous work, we documented every case in which a leader was convicted by a permanent, domestic civilian court.4 Over the past few decades, such convictions — of presidents, prime ministers, and dictators — have become increasingly common. Many leaders have been tried and convicted for corruption, human-rights violations, and acts of political violence, yet sentences for subverting democracy remain rare. Since 1946, only eleven former heads of government worldwide have been convicted for coups or self-coups.5 The three most recent ones all took place in Latin America — Bolsonaro included.
Bolsonaro’s conviction was also comparatively quick, coming just a few years after the attempted coup. In most cases, convictions have taken much longer, and were delivered only after the regime that resulted from the coup had collapsed. Apart from Bolsonaro, only Azerbaijan’s Surat Huseynov and Peru’s Pedro Castillo have been convicted for plotting a coup that eventually failed. Yet Azerbaijan in 1994 was not highly democratic, and Huseynov had already been out of power since the preceding year, making that case a coup attempt rather than a self-coup. This makes Bolsonaro’s conviction possibly the first example of a former leader who was convicted for plotting a self-coup that eventually failed in a democratic regime; Castillo’s conviction is the second, handed down just two months after Bolsonaro’s.
Of course, several leaders who subverted democracy were never convicted for their actions. The recent examples in Tunisia and El Salvador echo episodes from the 1990s, when democratically elected Boris Yeltsin of newly independent Russia and Alberto Fujimori of Peru showed how such gambits could succeed. In 1993, Yeltsin unilaterally dissolved the Russian parliament. When the Constitutional Court deemed the move illegal and legislators voted to impeach him, he ordered the army to storm the parliament building. Yeltsin then ruled by decree until a new constitution — vastly expanding presidential powers — was approved three months later. A year earlier, Fujimori dissolved Congress, suspended the constitution, and established an authoritarian regime that endured until 2000. He was eventually convicted in 2009 for political violence and corruption, though not for dismantling democracy. Yeltsin, by contrast, faced no legal consequences.
Even failed self-coups often fade without consequences in the courts. In Guatemala (1993), President Jorge Serrano Elías tried to dissolve Congress, suspend the constitution, and rule by decree, only to flee the country after the military withdrew support; he was never prosecuted. In Indonesia (2001), President Abdurrahman Wahid declared a state of emergency and sought to dissolve parliament by decree. Although he was swiftly impeached, he was never brought to trial. In Ecuador (2005), President Lucio Gutiérrez attempted to pack the courts and govern unilaterally, until mass protests forced his removal — again, without legal consequences. As these cases illustrate, impeachment and swift removal from office have been far more common institutional responses than judicial convictions.
And even when convictions occur, they do not always endure. Successors may later issue pardons, as in South Korea, where Chun Doo-hwan and Roh Tae-woo were pardoned just a year after their convictions for the 1979 coup. The same happened, after a longer interval, with Huseynov, who was convicted by Azerbaijani courts in 1999 and pardoned in 2004. In rarer instances, courts themselves may overturn verdicts on appeal, as in the cases of Pakistan’s Pervez Musharraf and Bolivia’s Jeanine Á~nez. So, while successful coups and self-coups seldom lead to punishment by the regimes they install, convictions for failed self-coups are also rare, even over time. Bolsonaro’s sentencing thus represents an exceptionally rare event, historically and comparatively — one that therefore calls to be unpacked and understood.
Bolsonaro’s Attempted Coup
Bolsonaro is frequently cast alongside other authoritarian populists who have followed a similar playbook for eroding democratic institutions. In fact, one could easily place many of his actions within Nancy Bermeo’s typology of contemporary varieties of democratic backsliding.6 While in office, Bolsonaro initially pursued a gradual strategy, comprising attempts at executive aggrandizement and strategic electoral manipulation, many of which were met with resistance from the legislature and the courts. This included measures such as militarization of the civilian bureaucracy, weaponization of the national-security law, and proposals to introduce printed ballots in a country long used to electronic voting machines.7
When this gradual route failed, Bolsonaro attempted a more abrupt one — similar to what Bermeo calls a “promissory coup” — which eventually got him convicted. As often happens, his strategy hinged on framing authoritarian rupture as a corrective measure to “restore democracy.”8 The premise had been laid during the four years leading up to Brazil’s October 2022 presidential election. Throughout his presidency Bolsonaro repeatedly questioned the reliability of the electronic voting system and preemptively delegitimized electoral authorities, particularly the TSE. Once he lost the election by the tightest margin (49.1 percent to Luiz Inácio Lula da Silva’s 50.9 percent) since redemocratization, he refused to enable a peaceful transfer of power. The coup plan he presented to the heads of the armed forces envisioned declaring a state of emergency, annulling the election results, and replacing the TSE with a new body staffed by top military personnel who would oversee fresh elections that would presumably result in Bolsonaro’s victory this time around.9
The scheme also included violent escalatory measures: plans to assassinate the newly elected Lula and his vice-president, as well as the Supreme Court justice who headed the TSE during the elections and had overseen criminal investigations into Bolsonaro and his acolytes. Beyond all this, almost immediately after the election results were made public, nationwide protests broke out and roadblocks were set up by Bolsonaro supporters who questioned the electoral results. In the month preceding the transfer of power, as encampments near military installations persisted, some supporters even attempted to storm the headquarters of the Federal Police and to plant a bomb in Brasília’s airport.10
Ultimately, as the new government took office, supporters of the outgoing president launched coordinated attacks on federal buildings on 8 January 2023. These assaults were designed as a “last attempt” to generate chaos and justify the declaration of a state of emergency that would place political authority in the hands of the armed forces, including some who were still loyal to Bolsonaro. Far from a spontaneous riot, the events exposed Bolsonaro’s long-running two-track approach: incremental institutional capture paired with the ever-present threat of a self-coup and the latent prospect of a coup d’état after his departure from office.
The investigations into Bolsonaro’s coup attempt gained speed and depth after Lula’s inauguration on 1 January 2023. They were conducted by the Federal Police under the stewardship of Justice Alexandre de Moraes of the Federal Supreme Court, who had already handled several cases involving Bolsonaro and his allies. With the appointment of a new prosecutor-general in December 2023, the inquiries gained further momentum, uncovering extensive communications and planning networks among Bolsonaro’s inner circle.
A crucial breakthrough came when his aide-de-camp, Lieutenant-Colonel Mauro Cid, signed a plea-bargain deal in late 2023. Cid’s testimony, along with documents retrieved from his phone and computer, exposed detailed plans to annul the election and enlist the military to prevent the transfer of power. Investigators also revealed a wide array of corroborating material that linked the former president and his inner circle to the coup attempt: financial records revealing the allocation of resources to coordinate coup-related activities, such as setting up pro-coup encampments near military installations; drafts found in Bolsonaro’s files of a decree and speech declaring the annulment of the 2022 election results; meeting records and intercepted communications showing the former president’s and co-conspirators’ repeated attempts to pressure the heads of the armed forces into endorsing the plan; and wiretaps and encrypted messages among officers sympathetic to Bolsonaro that discussed the logistics, timing, and execution of an operation to kidnap Justice Moraes.
Most strikingly, the heads of the army and air force later testified that Bolsonaro had discussed the self-coup scenario with them and sought their support — admissions that lent decisive credibility to the prosecution’s case. It was, in fact, largely due to lack of support from within military command that the plot failed to materialize; only the head of the navy embraced the plan, for which he was eventually convicted. Together, these findings exposed a coordinated effort — among both civilian and military officials — to subvert the constitutional order.11
Given the involvement of a sitting president at the time of the offense, the case was tried directly at the Federal Supreme Court. Bolsonaro was indicted by the prosecutor-general in March 2025. Fearing he would flee, the Court ordered him to wear an electronic ankle tag and prohibited him from using social media starting in July. He soon violated these measures and was put on pretrial house arrest in August. The trial took place the following month before one of the STF’s five-justice panels. On most counts, the vote was 4 to 1 to convict. Bolsonaro was convicted for an attempted coup d’état, attempted violent abolition of the democratic rule of law, participation in an armed criminal organization, qualified damage, and deterioration of protected heritage property. Convicted alongside Bolsonaro were his 2022 running mate, Walter Braga Netto, who had previously served as defense minister and chief of staff; a former army commander and defense minister; a former justice minister; a former navy commander; a former intelligence chief; and a retired general who had served in a cabinet-level position similar to national security adviser.
Bolsonaro’s sentencing was possible only because the courts had withstood his attacks while he was in office and because he was no longer president when the trial began. Had he remained in power or wielded his presidential powers unchecked, a conviction would have been extremely unlikely. More than just sentencing Bolsonaro, the high courts shaped the very conditions which made that outcome possible. They did so by constraining Bolsonaro’s most authoritarian maneuvers and enabling his opponents to eventually defeat him outside the courtroom.
Constraining Bolsonaro and His Allies
Before entertaining an open coup, Bolsonaro had seen many parts of his more incremental strategy blocked by the STF and the TSE, making him increasingly belligerent. But unlike in countries where authoritarian populists have succeeded, Bolsonaro was unable to curb the courts’ powers, as Congress never went along with the president’s proposals to change the composition of the courts. Institutional design also played a significant role in this outcome. The STF wields unusually broad powers for a supreme court: It oversees criminal proceedings against high-ranking officials and rules on a vast docket that keeps the STF at the center of political life. Under Brazil’s constitution, even small parties and civil society organizations can directly petition the Court to review executive and legislative measures, greatly broadening access to judicial oversight. The STF’s reach also extends through the TSE, where three of seven justices — including the court’s president — are drawn from the STF, ensuring close coordination between constitutional and electoral oversight. This combination of insulation, investigative authority, and overlapping jurisdiction helped Brazil’s higher courts both to withstand sustained attacks from the executive and to respond to them, despite a general prosecutor who was predominantly loyal to Bolsonaro.
A core component of the high courts’ response, therefore, was constraining — imposing direct costs on the president, his administration, and his allies through rulings that directly limited executive attempts to concentrate power. Using its power of judicial review, the STF rapidly dismissed several measures taken by the administration that aimed at executive aggrandizement. These included rulings that suspended or limited the scope of presidential decrees reducing civil society participation in public-policy councils, the appointment of a new Federal Police director amid allegations of presidential interference with ongoing investigations into one of Bolsonaro’s sons, and a decree shielding public officials for decisions taken during the covid-19 pandemic.12 Also, the Court helped to limit the Federal Police’s weaponization of the National Security Law (which was inherited from the military regime) to investigate journalists and opponents by publicly denouncing these moves and handing down decisions that pushed back against them.
Perhaps even more consequential — and controversial — was the STF’s reliance on its criminal jurisdiction. In a move that risked blurring the line between judicial and investigative functions, beginning in March 2019, the Court launched its own criminal investigations, for the most part bypassing the Bolsonaro-appointed prosecutor-general. Drawing on internal, never-before-invoked rules that authorize self-initiated inquiries into crimes committed inside the courtroom, the STF opened a series of high-profile investigations into fake news, digital militias, and antidemocratic acts allegedly undertaken by Bolsonaro and his allies targeting the Court and its justices. The Court’s supporters viewed these self-initiated inquiries as essential to defending judicial independence amid escalating attacks, while critics warned that they concentrated excessive power in the hands of the justices and risked undermining due-process norms. The episode captured the core dilemma facing Brazil’s courts: how to uphold democratic values without compromising the principles on which they rest.
Through these inquiries, Bolsonaro was placed under criminal investigation while still in office. Justice Moraes ordered arrests, froze bank accounts, lifted financial, phone, and digital privacy for investigative purposes, and blocked the social-media accounts of some of the president’s allies. A few of these cases led to true standoffs between the STF and the president. In February 2021, Moraes ordered the arrest of Congressmen Daniel Silveira after he posted a video attacking STF justices, a decision that was later upheld both by the full court and by Congress. Paradoxically, the ruling was based on the same National Security Law which the Federal Police under Bolsonaro had weaponized against his opponents. This time, however, the Bolsonaro-appointed prosecutor-general joined in, filing an indictment that eventually led to Silveira’s conviction in April 2022. Bolsonaro responded by quickly pardoning Silveira.13 The STF, however, froze Silveira’s bank accounts. The stalemate would only be resolved after Bolsonaro was no longer president. In February 2023, when Silveira too was out of office, the STF again ordered his arrest and, in May, annulled the presidential pardon.
The TSE, for its part, focused on constraining efforts by the president and his supporters to undermine electoral integrity. It imposed measures to curb misinformation about Brazil’s electronic voting system and sanctioned social-media platforms that failed to comply with court orders to remove content. In October 2021, the TSE set a precedent by removing State Representative Fernando Francischini, a Bolsonaro ally, from office for making false claims about the voting system and barred him from holding office for eight years. This precedent informed the TSE’s June 2023 decision that declared Bolsonaro ineligible until 2030 for similar reasons. The Court ruled that the former president had abused political power and misused state media by holding an official meeting with foreign ambassadors, which was broadcast by state media, where he repeated baseless claims about the unreliability of the electronic voting system.
Also, during the 2022 elections, when Justice Moraes was serving as its president due to seniority rules, the TSE enacted new regulations requiring social-media platforms to remove online content that violated court rulings within two hours; failure to comply risked suspension of services in Brazil.14 The Court’s assertive stance continued on election day, when the Federal Highway Police carried out coordinated roadblocks and vehicle inspections in Lula’s strongholds to hinder voters’ access to the polls. The TSE ordered them to stop; when news came that the police did not back down, Moraes threatened to arrest the Highway Police chief, who then complied. Lula’s legal team even petitioned for a two-hour extension of voting, but Moraes denied the request to avoid fueling accusations of tilting the playing field to Lula’s benefit, reportedly remarking that such a delay “was all the president wanted.”15 Then, after election results were made public, Bolsonaro’s party filed a request for the TSE to invalidate votes cast in roughly 60 percent of the voting machines, claiming that they were unreliable — the TSE quickly dismissed the claim and fined the party.
Enabling Bolsonaro’s Opponents
In addition to curbing presidential overreach directly, the courts also acted indirectly by empowering other officials and politicians to compete with and impose costs on the president and his allies.16 This less confrontational dimension of the reaction was no less effective: By bolstering the authority of governors, legislators, and political rivals, the judiciary diffused power away from the presidency without escalating direct conflict with it. Through this approach — and three rulings, in particular — the STF enabled others to impose tangible political costs on Bolsonaro.
The first addressed a presidential decree granting the federal executive exclusive authority over Brazil’s covid policy. Responding to a case filed by a small opposition party, the STF struck down the decree and affirmed that states and municipalities retained authority to impose stricter measures such as masking and quarantine mandates. The decision decentralized authority and opened space for subnational leaders — most prominently, S~ao Paulo’s governor Jo~ao Doria, a former ally who broke with Bolsonaro over his handling of the pandemic — to outpace the federal government in visibility and effectiveness on vaccination and pandemic management.
The second ruling compelled Senate president Rodrigo Pacheco to open a parliamentary inquiry into the government’s pandemic response, despite his reluctance and despite the risk of intruding on the legislature’s internal affairs. The decision came at the height of the crisis, as daily deaths in Brazil exceeded four thousand. The ensuing investigation exposed negligence, corruption, and misconduct involving senior officials in Bolsonaro’s inner circle and gave legislators a powerful platform to challenge the president’s actions in nationally televised hearings. In this sense, the judiciary served as an institution that, as Jeffrey Staton and colleagues note, facilitates information exchange between governments and their opponents, helping the opposition to check executive overreach.17 By the time the hearings concluded later that year, Bolsonaro’s approval ratings had plunged, indicating that judicial intervention strengthened his rivals in Congress and, in turn, contributed to the erosion of his political standing.
The STF’s most consequential move came when it overturned Lula’s corruption convictions stemming from the Car Wash (Lava Jato) scandal. This cleared the way for his presidential candidacy in 2022. Those convictions, which had been upheld by multiple courts and ratified by both the STF and the TSE, had disqualified Lula from running in 2018. Yet as evidence of misconduct within the Car Wash investigation mounted, including revelations that Curitiba judge Sérgio Moro had collaborated with prosecutors, the STF reversed its stance. First, it ruled that Moro’s court lacked jurisdiction, as the alleged crimes should have been tried in Brasília; second, it found that Moro had demonstrated bias by collaborating strategically with the case’s prosecutors and then becoming Bolsonaro’s justice minister less than two years after Lula’s conviction. By overturning Lula’s conviction, the STF not only restored his political rights but also reintroduced a credible challenger capable of uniting the opposition.
The high courts also worked to empower other institutions by pressing Congress to resist Bolsonaro’s attempts to erode confidence in the electoral process. Justices lobbied legislators to block proposals to reintroduce paper ballots, a central item on Bolsonaro’s agenda, and publicly defended the reliability of Brazil’s long-standing electronic voting system. Congress followed suit, rejecting Bolsonaro’s bill in September 2021. Efforts to enlist the military, however, proved less effective. Although the courts invited the armed forces to join a commission on electoral integrity, the military largely echoed Bolsonaro’s talking points or remained silent in the face of his attacks on the system’s credibility.
At the same time, however, the high courts did not seize every opportunity to strengthen those positioned to counter Bolsonaro. In late 2020, for example, when center-right moderate Rodrigo Maia sought reelection as president of the Chamber of Deputies, the STF blocked his bid. Although the Constitution forbids reelection to that post within the same legislative term, the Court’s ruling was divided and could easily have gone the other way. This decision cleared the path for Arthur Lira, a congressional leader closely aligned with Bolsonaro, to assume the Chamber’s presidency, a position that is crucial to block impeachment proceedings, thus emboldening Bolsonaro to intensify his attacks on the courts.
Avoiding Confrontation with Congress and Other Trade-Offs
In balancing responses to Bolsonaro’s attacks with efforts both to constrain him and to empower others, Brazil’s high courts pursued a strategy of institutional self-preservation. These outcomes were far from inevitable. Before Bolsonaro’s presidency, the STF was deeply fragmented, often described as “eleven islands,” reflecting internal divisions among its eleven members and its weak institutional coordination.18 Confronted with a direct threat, however, the justices closed ranks. Judiciaries, when striving to defend themselves while helping to preserve democracy, will confront certain dilemmas. Knowing that its approach came with substantial risks and trade-offs, the Court had to act surgically.
At the core of its strategy was an effort to avoid direct clashes with Congress through a more accommodating stance on anticorruption.19 The rollback of the decade-long Car Wash investigations, which had implicated dozens of members of Congress and risked incriminating many more, marked a considerable retreat from the judiciary’s more assertive stance before Bolsonaro’s rise. By easing pressure on politicians across the spectrum, the STF secured broad political backing from a wide range of legislators who might otherwise have opposed it.
The same caution shaped the Court’s reaction to congressional budgetary maneuvers. The “secret budget” scandal, which diverted large discretionary resources to lawmakers with little transparency or oversight, empowered Congress. Yet the STF chose not to intervene — and did so, even if to small success, only after Bolsonaro left office. By giving legislators what they wanted, the judiciary won near carte blanche to defend itself against Bolsonaro’s attacks.
This accommodation also opened space for pivotal segments of Congress, wavering between support for the president and the opposition, to take a more assertive stance against his authoritarian drift. As often occurs, assertiveness coexisted with complicity: Some lawmakers advanced democratic safeguards even as they benefited from the “secret budget” and from leniency on corruption. In 2021, as it became clear that federal agencies were relying on the National Security Law inherited from the military regime to investigate Bolsonaro’s opponents, STF justice Gilmar Mendes warned the government not to use the law for such purposes, arguing that doing so would violate constitutional guarantees of free expression. A key contradiction, which may help to explain why the Court refrained from deciding on the case, was that it too had been invoking the National Security Law in many of its rulings to constrain Bolsonaro and his most radical allies, such as Congressmen Silveira.
Repealing the old legislation, Congress passed a law defining “crimes against democracy,” including violent attempts to abolish the rule of law or carry out coups d’état, while also establishing explicit protections for free expression and protest.20 Despite its relative weakness, congressional opposition forged alliances and seized this key opportunity. Only deputies from Bolsonaro’s party and from a few other small right-wing parties voted against the measure, which was approved with broad support despite about a sixth of deputies abstaining. Although signed by Bolsonaro himself, the law failed to deter the president and his inner circle from committing such offenses. Later, however, it would provide the key legal basis for their prosecution and conviction.
Another sign of the STF’s caution was its reluctance to confront Bolsonaro’s popular base. Seeking to avoid alienating allies of the president, the judiciary often sidestepped issues that held broad public appeal. It refrained, for example, from intervening on irregularities surrounding the massive cash-transfer program Auxílio Brasil, an expansive new fuel-subsidy program, and other ad hoc public benefits introduced to bolster Bolsonaro’s electoral prospects. Steering clear of these politically sensitive measures reduced the risk of open conflict with the executive but also signaled the limits of judicial assertiveness when popular programs were at stake.
The TSE was far less restrained when it came to defending the electoral system. As the president sought to discredit electronic voting and sow doubts about the 2022 election, its justices issued forceful public statements warning that such assaults would have consequences. Coupled with several constraining measures adopted by the TSE, these interventions, aimed at domestic and international audiences alike, signaled that the courts would not stand by while the democratic process was undermined. Through this extrajudicial communication, the justices sought to shape public opinion and deter antidemocratic behavior. Still, on many occasions, their outspokenness blurred the line between independence and political engagement, reinforcing perceptions of judicial politicization.21 On election day, the TSE took decisive action, coordinating closely with congressional leaders and foreign governments to affirm the electoral results and ensure the transfer of power.
Lessons for Democracies Under Threat
Empirical research on the key determinants of democratic backsliding shows that when national courts are independent, democracy is less likely to experience decline.22 Judicial behavior is thus at the heart of democratic resilience — how courts respond when confronted with executive threats and the trade-offs those responses entail. The concrete choices that judges make, and their political consequences, help to explain why some democracies withstand authoritarian pressures while others do not.23
Brazil under Jair Bolsonaro offers a vivid illustration. From the outset, Bolsonaro sought to weaken judicial constraints on his government and discredit the electoral system. His court-bashing in speeches and at rallies mobilized his base and allowed him to pose as an antiestablishment crusader against what he claimed was a “corrupt” judiciary, even as a sitting president who had served seven terms in Congress. Amid these attacks, the prevailing mood within the higher courts was perhaps best captured by one STF justice’s comment in November 2022, a month after Bolsonaro’s electoral defeat: “We have not slept since February 2019.”24 The remark conveyed both the vigilance required to contain threats received and the exhaustion of nearly four years of unrelenting institutional confrontation.
Alongside the high courts’ resolve and institutional design, including its wide-ranging jurisdiction, Bolsonaro’s efforts to subvert democracy were limited by Brazil’s fragmented party system and bicameral legislature, which denied him a stable congressional majority.25 At the same time, the courts played a pivotal role not only in containing Bolsonaro’s initiatives by themselves, but also in further empowering congressional minorities and political contenders, all while avoiding conflicts that could alienate their own congressional support.
The courts, as a result, did not stop Bolsonaro on their own. Safeguarding democracy depends on coordination among political contenders, the media, civil society, and even foreign governments. It would be na¦ve to imagine that a single court could contain a president backed by large segments of the population and the armed forces. Acting in self-defense, Brazil’s high courts forged informal alliances, empowered other institutions, and avoided direct confrontation with congressional leaders, even those aligned with Bolsonaro. By steering clear of open clashes, the judiciary preserved its authority while helping to shift legislators’ incentives toward defending democratic norms rather than enabling their erosion. One key initiative was passing the law on “crimes against democracy,” which was later used to convict Bolsonaro. Another was the 2021 defeat of Bolsonaro’s bill to introduce paper ballots. Yet another was the swift recognition of the 2022 electoral results by congressional leadership. Although these crucial developments involved multiple players, the Courts’ support was decisive.
Despite all these efforts, Bolsonaro lost by only a razor-thin margin. His narrow defeat reflected the success of a broad, cross-ideological electoral coalition that united figures from the left and right and was supported by a wide spectrum of Brazilian civil society. The opposition’s victory also rested on meticulously crafted political alliances at the state level and among legislative candidates, reinforced by credible promises of power sharing. Together, these efforts raised the costs of repression and constrained the armed forces’ room for intervention, ultimately helping to ensure the transfer of power.
Lula’s inauguration and the subsequent investigations under renewed democratic conditions created the political and institutional space necessary to hold Bolsonaro accountable before the law. Yet the aftermath of his conviction also reveals the tensions that remain. So far, Bolsonaro’s political ineligibility and conviction have not put an end to his movement. One of Bolsonaro’s sons, for example, allegedly worked closely with the sitting U.S. administration to pressure the STF to acquit Bolsonaro.26 Ahead of the September 2025 trial, the U.S. government put 50 percent tariffs on Brazil, revoked visas for Tribunal justices, and imposed Global Magnitsky sanctions on Justice Moraes. After the trial, the sanctions were extended to Moraes’s wife. Yet not only did the verdict not change, but the STF is currently considering taking other criminal cases against Bolsonaro, including investigations on attempts by the former president to interfere with the Federal Police and his management of the pandemic.
Although Lula is unlikely to grant Bolsonaro a pardon, Congress could still pass a sweeping amnesty that extends to the former president, his allies, and participants in the January 8 insurrection. So far, efforts to do so have failed, but the 2026 elections — when two-thirds of the Senate seats will be contested — could shift this balance. A new president or Senate majority sympathetic to Bolsonaro might seek to pardon him and pressure the STF by threatening to impeach justices or limit the court’s powers. In fact, impeachment requests against STF justices reached record levels during Bolsonaro’s presidency. So far, the Senate has provided an essential legislative shield for the Court.
Brazil’s experience underscores enduring dilemmas faced by democracies confronting autocratizing leaders.27 The broader lesson is clear: Courts cannot wait until a populist authoritarian leaves office to act — by then, it may be too late. Yet effectiveness requires balancing assertiveness with restraint, choosing battles wisely rather than going full throttle. It also requires acknowledging, somewhat obviously, that courts cannot be the only bulwarks against democratic backsliding: To take action, the courts need support from defectors within the governing coalition, opposition politicians, media, and civil society groups; the courts’ actions, in turn, may empower these same groups to impose costs on autocratic leaders.
Still, it must be noted that although the Brazilian high courts did help to stop Bolsonaro, many of its decisions were unorthodox vis-`a-vis due-process standards. This raises several important questions: How far should courts go in defending democracy, and what safeguards would prevent judicial overreach? Should leaders who attack democratic institutions be punished to ensure accountability, or spared to preserve stability? Similar questions now test other democracies. In late 2024, Romania’s Constitutional Court took the extraordinary step of annulling the presidential election on grounds of foreign interference. And in South Korea, former president Yoon Suk Yeol is being tried for declaring martial law in December 2024 to bypass democratic institutions. In each case, courts are grappling with the same fundamental challenge — how to uphold the rule of law without tipping fragile political orders into deeper crisis.
NOTES
1.John Chin and Joseph Wright, “What Is a Self-Coup? South Korea President’s Attempt Ended in Failure — A Notable Exception in a Growing Global Trend,” The Conversation, 5 December 2024, https://theconversation.com/what-is-a-self-coup-south-korea-presidents-attempt-ended-in-failure-a-notable-exception-in-a-growing-global-trend-235738.
2. Nic Cheeseman et al., “How to Strengthen Democratic Resilience,” European Democracy Hub (December 2024), https://europeandemocracyhub.epd.eu/wp-content/uploads/2024/12/How-to-Strengthen-Democratic-Resilience-Five-Lessons-for-Democratic-Renewal_final.pdf
3. Carlos Fico, Utopia Autoritária Brasileira:Como os militares ameaçam a democracia brasileira desde o nascimento da República até hoje (São Paulo: Editora do Planeta, 2025).
4. Luciano Da Ros and Manoel Gehrke, “Convicting Politicians for Corruption: The Politics of Criminal Accountability,” Government and Opposition 59, no. 3 (2024): 951–75.
5. Jeanine Áñez’s case is more complex: She took power after a confrontation between the Morales government and Bolivia’s security forces that forced the president’s resignation; see Linda Farthing and Thomas Becker, Coup: A Story of Violence and Resistance in Bolivia (Chicago: Haymarket, 2021). Áñez and several former commanders were later convicted of “dereliction of duty” and “violating the constitution.” In 2025, the Bolivian Supreme Court overturned the ruling, leading to Áñez’s release from prison.
6. Nancy Bermeo, “On Democratic Backsliding,” Journal of Democracy 27 (January 2016): 5–19.
7. Michael W. Bauer et al., “Bureaucratic Militarization as a Mode of Democratic Backsliding: Lessons from Brazil.” Democratization 32, no. 3 (2025): 595–613; and Augusto N. C. Oliveira, “Democracy, Populism and the Printed Ballot Discourse,” Dados 67, no. 4 (2025): e20220148.
8. Bermeo, “On Democratic Backsliding,” 8.
9. Isadora Peron et al, “Brazil’s Top Court Advances New Cases over Coup Attempt,” Valor Econômico, 23 April 2025, https://valorinternational.globo.com/politics/news/2025/04/23/brazils-top-court-advances-new-cases-over-coup-attempt.ghtml.
10. Andre Romani and Gabriel Stargardter, “Bolsonaro’s Call to Arms Inspired Foiled Brazil Bomb Plot, Police Are Told,” Reuters, 26 December 2022, https://www.reuters.com/world/americas/bolsonaros-call-arms-inspired-foiled-brazil-bomb-plot-police-are-told-2022-12-26/
11. Naiara Galarraga Gortázar, “Brazil vs. Bolsonaro: Keys to a Historic Coup Attempt Trial,” El País, 9 September 2025, https://english.elpais.com/international/2025-09-09/brazil-vs-bolsonaro-keys-to-a-historic-coup-attempt-trial.html.
12. Oscar Vieira et al., “Supremocracy and Autocratic Infralegalism: The Behavior of the Brazilian Supreme Court During Bolsonaro’s Government,” Novos Estudos CEBRAP 41 (2024): 591–605.
13. Diego Wernick Arguelhes, “Public Opinion, Criminal Procedures, and Legislative Shields: How Supreme Court Judges Have Checked President Jair Bolsonaro in Brazil,” Georgetown Journal of International Affairs, 25 April 2022, https://gjia.georgetown.edu/2022/04/25/public-opinion-criminal-procedures-and-legislative-shields-howsupreme-court-judges-have-checked-president-jair-bolsonaro-in-brazil/
14. Clara I. Keller and Diego W. Arguelhes, “Facing Disinformation in Democratic Backsliding: The Role of Courts in Brazil,” VRÜ Verfassung und Recht in Übersee 57, no. 2 (2025): 187–209.
15. Felipe Recondo and Luiz Weber, O Tribunal: Como o Supremo se Uniu Ante a Ameaça Autoritária (São Paulo: Companhia das Letras, 2023).
16. For similar arguments about courts’ role in enabling actors in other contexts, see Benjamin Garcia-Holgado, “Overruling the Executive: Judicial Strategies to Resist Democratic Erosion,” Journal of Law and Courts 13, no. 1 (2025): 274–303.
17. Jeffrey Staton et al., Can Courts Be Bulwarks of Democracy? Judges and the Politics of Prudence (Cambridge: Cambridge University Press, 2022).
18. Felipe Recondo and Luiz Weber, Os Onze: O STF, seus bastidores e suas crises (São Paulo: Companhia das Letras, 2019).
19. Luciano Da Ros and Matthew Taylor, “Bolsonaro and the Judiciary: Between Accommodation and Confrontation,” in Bruno Speck and Peter Birle, eds., Brazil Under Bolsonaro: How Endangered Is Democracy? (Berlin: Ibero-Amerikanisches Institut Preußischer Kulturbesitz, 2022): 44–54.
20. “Military Dictatorship Era National Security Law Is Repealed,” Conectas, 2 September 2021, https://conectas.org/en/noticias/military-dictatorship-era-national-security-law-is-repealed.
21. Matthew Taylor, “Brazil’s Rule-of-Law Regression,” Current History 123, no. 850 (2024): 43–49.
22. Vanessa A. Boese et al., “How Democracies Prevail: Democratic Resilience as a Two-Stage Process,” Democratization 28, no. 5 (2021): 885–907; Douglas Gibler and Kirk A. Randazzo, “Testing the Effects of Independent Judiciaries on the Likelihood of Democratic Backsliding,” American Journal of Political Science 55, no. 3 (2011): 696–709; Jeffrey Staton et al. Can Courts Be Bulwarks of Democracy?
23. For how courts in Latin America have helped to contain democratic backsliding, see Laura Gamboa et al., “Courts Against Backsliding: Lessons from Latin America,” Law & Policy٤٦, no. 4 (2024): 358–79; Diego Zambrano et al., “How Latin America’s Judges Are Defending Democracy,” Journal of Democracy ٣٥ (January 2024): 118–33.
24. Matheus Leitão, “A frase assustadora de um ministro do STF resume o governo Bolsonaro,” Veja, 25 November 2022, https://veja.abril.com.br/coluna/matheus-leitao/a–frase-assustadora-de-um-ministro-do-stf-resume-o-governo-bolsonaro.
25. Marcus André Melo and Carlos Pereira, Por que a democracia brasileira não morreu? (São Paulo: Companhia das Letras, 2024).
26. Joan Royo Gual, “Eduardo Bolsonaro, the Man Urging Trump to Punish Brazil,” 29 July 2025, https://english.elpais.com/international/2025-07-29/eduardo-bolsonaro-the-man-urging-trump-to-punish-brazil.html
27.Matías Bianchi, Nic Cheeseman, and Jennifer Cyr, “The Myth of Democratic Resilience,” Journal of Democracy 36 (July 2025): 33–46.
Copyright © 2026 National Endowment for Democracy and Johns Hopkins University Press
Image credit: SERGIO LIMA/AFP via Getty Images