It started with an item in the feed: on July 28, 2026, Keiko Fujimori took office as president of Peru. Victory on the fourth attempt, a quarter-century after her father left the country under threat of trial. A thread was sticking out of the news item, and there it was: Keiko took her oath under the 1993 constitution—the very document her father rammed through after dissolving his own parliament. The document did not save him: he died a convicted man. But the constitution lives on, 33 years old, and his daughter’s power rests on it.
An autogolpe—from the Spanish for “self-coup”—is that particular genre of coup in which a head of state dissolves not their opponents, but their own institutions. Modern history has enough of them to fill a shelf. Peru in 1992 and Russia in 1993 are pure specimens: the president uses the army to clear out his own parliament. Chile in 1973 does not fit the definition—the military dictatorship arrived through a coup—but the 1980 constitution grew from the same logic of “order matters more than procedure,” and the conversation would be incomplete without it. The logic is similar in every case: a broken old order, a charismatic figure with the army at his back, and the certainty that democracy gets in the way of saving the country. And then it gets interesting—the constitutions born from these coups have met different fates. One was tweaked for a new master; another lived to see its author’s daughter in the presidential chair.
On the evening of April 5, 1992, Alberto Fujimori went on air and announced that he was “temporarily” dissolving Congress and “reorganizing” the judiciary. A tank was brought up to the parliament building, while a group of senators who tried to convene a session were dispersed with tear gas. Decree 25418 handed all legislative power to the president and suspended much of the 1979 constitution.
The argument was familiar in Latin America: democracy was not coping. Peru really was at war then—the Maoist Shining Path controlled entire districts, the economy was collapsing, hyperinflation was devouring wages in weeks. Fujimori sold the coup as surgery: to cure the country, you had to disconnect it from politics for a while. The polls published at the time gave him monstrous support—APOYO reported that 71% approved of dissolving Congress and 89% of reorganizing the courts. This was, however, one of those cases where statistics and their interpretation arrived on the same truck: researchers openly called those figures “semantic grazing”—the government held a monopoly on how the question was framed.
A year later, on November 22, 1992, elections were held for the Democratic Constituent Congress—the body meant to write a new constitution. The opposition boycotted them, calling them a farce. The document was published on September 4, 1993, and put to a referendum.
On October 31, 1993, Peruvians answered one question: did they approve the constitution written by that congress?
The result was not a rout, but suspiciously narrow. In favor—3,895,763 votes, 52.33%. Against—3,548,334, or 47.67%. The gap was under five percentage points: just over 347,000 votes out of eight million cast.
And this is where the central paradox kicks in. A document backed by barely half of those who showed up, while half the country did not show up at all (turnout was 70.4%), became the country’s basic law for three decades—including for the people who cast 47% of the vote against it.
What was written into that text matters too. The new constitution limited the president to two five-year terms, made parliament unicameral, allowed amendments either by referendum or by a two-thirds vote of two successive congresses, and permitted referendums by popular initiative with signatures from just 0.3% of registered voters. It looks like a set of safeguards. The problem was that the author fitted them with an off switch.
Just three years later, Fujimori ran into his own limit. Article 112 allowed reelection for only one additional consecutive term—meaning that after his 1990 victory and reelection in 1995, a third straight run in 2000 was barred. The limit itself had been written in that fresh post-coup moment, when term limits looked like protection against another “eternal” ruler.
It was bypassed not with a tank, but with a circular. In 1996, the Fujimorist Congress passed the “Law of Authentic Interpretation” (Ley 26 657): since Fujimori’s first term had begun in 1990, before 1993, that term “did not count,” and in 2000 the president was, de facto, running for only a second term. Legally, it was a fiction—the constitutional limit had been written precisely with the man now evading it in mind. But formally, the cork was out, and Fujimori ran for a third term and won it.
Then came the thing the charter’s author had not planned for. In 2000, videos surfaced: intelligence chief Vladimiro Montesinos handing out money to members of Congress. The regime crumbled in weeks. Fujimori flew to Japan—the country of his parents—and from there, in a gesture emblematic of the dial-up-and-fax era, sent in his resignation by fax. Later he returned to Peru, received 25 years in cases involving murders and kidnappings, and separate sentences for embezzlement and corruption. In December 2017, he was pardoned on health grounds, but he was not released until December 2023, after long court battles: the pardon had to be defended twice. He died in September 2024, nine months after his release, still convicted but already free.
That is where the oddity hides. The limit Fujimori wrote in for himself was supposed to stop him—and did not: he bypassed it with a circular. He fell to corruption, while the constitution outlived its author, his regime, and everyone who later tried in vain to tear it down. The author was gone. The charter remained.
The maneuver Fujimori used has a public shadow—and it is not about the referendum being bad.
In 2018, President Martín Vizcarra, who had come to power after a string of corruption scandals, proposed his own package of constitutional reforms and put four questions to a referendum. Three passed by crushing margins—85–87% “in favor.” But the fourth, on restoring a bicameral parliament, was rejected by an equally mirror-image tally: 90.5% “against.”
Six years passed. In March 2024, Congress—the very unicameral body that had grown out of the 1993 charter—voted to bring back the Senate just like that, without a referendum: 91 votes for, 30 against, one abstention. Under the new arrangement, parliament will consist of 130 representatives and at least 60 senators, and it will begin operating after the 2026 elections—the very elections that brought Keiko Fujimori the presidency. At the same sitting, lawmakers removed the limit that had barred election to two consecutive terms.
The only time the country rejected part of the 1993 charter by popular vote was in 2018, on bicameralism. Six years later, Congress overturned that decision in an ordinary session. The document changes when the people it is meant to constrain want it to.
To understand why the fate of Peru’s charter is not a universal rule, it helps to look at its relatives.
Russia’s charter was born from the same script. On September 21, 1993, Boris Yeltsin dissolved the Supreme Soviet; on October 4, tanks shelled the White House, and the reformist president replaced the Soviet system with a presidential one he had dictated himself. In the December 12 referendum, it was approved by 58.43% on turnout of 54.37%. Then the paths diverged: in 2020, Russia’s was rewritten through amendments that reset the sitting president’s term count.
Chile’s is sturdier than either. It was written under Pinochet and likewise forced through a plebiscite, and in 2022 the country tried to replace it and rejected the proposed text—61.9% against to 38.1% in favor. In 2023 there was a different draft, and it was rejected more gently, but no less decisively: 55.8% against. After two rejections, Chile remained with the 1980 charter—it has been extensively rewritten since 2005, removing the last “anti-democratic enclaves,” such as appointed senators, but leaving the framework itself untouched. Pinochet died in 2006 while under investigation and house arrest; his constitution survives to this day.
Three charters, three coups, three countries—and none collapsed because it had been born ugly. Dictators and presidents fell, but the texts remained: they were not repealed, they were filed down to fit new conditions. The only thing that truly does not outlive a constitution’s authorial edition is the author himself.
Now the circle has closed. Keiko Fujimori, leader of the Popular Force party, took power on July 28, 2026, under that very 1993 constitution—the one her father laid down with a tank on the steps of Congress.
There is a temptation to read this as a dynastic triumph: the family returned, the document held. But it is more honest to read it differently. The constitution outlived its author not because it was good or just, but because nobody assembled enough strength to replace it. Its opponents were just under half, while more than half was needed. Its successors had no desire to: it was more profitable to tweak the limits than to build a new document. For 33 years, Peru did not answer whether a constitution can be founded on the dispersal of parliament. In July 2026, the country simply brought a person named Fujimori to power and left the 1993 charter to keep working.