An annulled election is rare enough to be shocking and common enough that any results archive has to decide how to record one. Three recent European cases sit on this site, and they are usefully different from each other.
Romania 2024: a first round cancelled outright
The first round of Romania's presidential election, held on 24 November 2024, produced a result nobody had forecast. Călin Georgescu, an ultranationalist independent who had barely registered in opinion polling, finished first with 22.94%, narrowly ahead of the reformist USR leader Elena Lasconi on 19.18%.
Before the run-off could be held, the Constitutional Court took the unprecedented step of cancelling the entire election, citing declassified intelligence describing a coordinated, undeclared social-media campaign on Georgescu's behalf. The election was re-run from the beginning in May 2025, with Georgescu barred from standing; Nicușor Dan won the run-off with 53.60% against George Simion's 46.40%.
What makes this case distinctive is that nothing was alleged to be wrong with the count. The votes were tallied correctly. The annulment concerned the conditions under which the votes were cast — which is a far more contested basis for voiding an election, and was argued as such at the time.
Austria 2016: a run-off annulled over procedure
Austria's 2016 presidential run-off was extremely close, and the Freedom Party challenged it. The Constitutional Court annulled the result — not because it found that any vote had been altered, but because it found that the rules for handling absentee ballots had not been followed: envelopes opened early, counting begun before authorised officials were present.
The court's reasoning is worth understanding, because it is often misreported as "the election was rigged". It was not. The finding was that procedural safeguards had been breached widely enough that the result could have been affected, and that in a contest this close the safeguards themselves are what make the outcome legitimate. The re-run in December returned Alexander Van der Bellen with 53.79% to Norbert Hofer's 46.21% — a wider margin than the annulled vote, and no evidence of fraud in either.
Germany 2021: a partial repeat
The third pattern is the least dramatic and probably the most useful. Berlin's administration of the 2021 federal election failed in specific, documented ways — missing ballot papers, wrong ballots issued, queues that kept polling stations open past the legal close. Rather than annul a national election, the courts ordered a partial repeat, held in 2024 in the affected Berlin districts only.
The rest of the country's votes stood. Only the compromised areas voted again, and the national seat totals were adjusted by the difference. It is a proportionate remedy, and it is only available in systems where the flaw can be localised — which is why it fits a constituency-based parliamentary election and would not fit a single national presidential count.
Why annulment is rarer than it feels
Reading three cases in a row makes courts look trigger-happy. They are not. Challenges to election results are filed constantly and almost all of them fail, because most courts apply a version of the same test: could the irregularity have changed the outcome?
That test does most of the filtering. A hundred miscounted ballots in a constituency decided by twenty thousand cannot have changed anything, so the remedy is a correction, not a re-run. The same hundred ballots in a race decided by fifty is a different matter. It is why annulments cluster in extremely close contests — Austria's run-off being the textbook example — and why a large margin is, in practice, its own legal defence.
Courts also have milder tools and reach for them first: ordering a recount, correcting a tally, re-running a single polling station, or referring individuals for prosecution while leaving the result standing. Voiding an election disenfranchises everyone who voted correctly, so it is a last resort rather than a first response — and a system that annulled easily would hand every losing candidate an incentive to manufacture procedural complaints.
How a results archive should record this
Annulled elections create a genuine editorial problem: the figures exist, they were officially published, and they are legally void. Deleting them is wrong — they happened, and the 2024 Romanian first round is a matter of historical record that a reader has every reason to look up. Presenting them as ordinary results is also wrong.
The approach here is to keep the page, publish the real figures, and label the status explicitly, with the reason and the date of the annulment in the page's own text. The annulled round and its re-run each get their own page at their own date, and each links to the other. A reader arriving from a search for "Romania 2024 presidential result" gets the number they were looking for, immediately followed by the fact that it was voided.
The same principle governs the quieter cases: a seat total that shifts after a partial repeat, a figure revised during certification, a referendum voided by a turnout quorum. In each case the honest record is the published number plus its status — never one without the other.