Moldova says yes to Europe by 10,564 votes. Sandu still has an election to win.

Moldova's European referendum was decided by a margin small enough to fit inside a modest town.

The Central Electoral Commission's count, approved on 25 October, recorded 749,719 votes in favour of the constitutional change and 739,155 against. The difference was 10,564 votes out of almost 1.49 million valid ballots: 50.35 percent for yes.

For President Maia Sandu, who had championed the measure, it was a victory with an immediate warning attached. Her European project survived the 20 October vote. Her own re-election was still unresolved.

On the separate presidential ballot held that day, Sandu received 42.49 percent, ahead of former prosecutor general Alexandr Stoianoglo on 25.95 percent. Neither had the majority needed to win outright. They would face each other again on 3 November.

The two results do not describe the same coalition of voters. Some citizens could support constitutional recognition of EU integration while preferring another president. Others could oppose the referendum's timing or wording without wanting Moldova to abandon all cooperation with Europe.

That distinction is now central to the runoff. Sandu must convert the narrow success of an institutional proposal into broader support for her continued leadership. Stoianoglo, backed by the Socialist Party, must attract voters beyond his first-round base while resisting her portrayal of the contest as a choice about Russian influence.

Both campaigns are operating amid serious allegations of organised vote-buying and external interference. Those allegations have prompted police investigations and international condemnation. They have also intensified a political argument in which almost every disputed result is being asked to carry the weight of Moldova's entire geopolitical future.

A late count changes the result

During the night after polling, the referendum appeared to be heading for defeat. Results from abroad, where Moldova's diaspora strongly favoured the European course, eventually pushed yes ahead. The first published percentages shifted as ballots were counted and the electoral authorities consolidated returns.

That progression is why the final approved totals matter more than a screenshot of an early count. Ballots arriving in the tally at different times do not necessarily represent the electorate in the same proportions. A change in the lead can reflect the geography of reporting rather than a change in the ballots themselves.

The margin is nevertheless politically stark. It gives the constitutional amendment a majority without supplying the broad consensus Sandu would have preferred. The referendum results have been sent to the Constitutional Court for the next stage of the process.

The vote concerned the constitutional basis for European integration. It was not an accession treaty, and it did not make Moldova an EU member. Membership negotiations had already been formally opened at the first accession conference on 25 June.

The constitutional step can strengthen the domestic foundation of that policy. It cannot bind EU governments to accept Moldova on a particular date or remove the reforms required by the accession process. The country's courts, public administration and economy still have work to do.

For supporters, anchoring the course in the constitution makes reversal more difficult than changing a cabinet programme. For critics, attaching the referendum to a presidential election allowed the incumbent to place her own campaign alongside a national cause with wider appeal.

The international observation mission addressed that concern without dismissing the election. Its preliminary assessment described professional administration and a competitive contest, while finding that the simultaneous campaigns, favourable coverage for the incumbent and misuse of public resources prevented fully equal conditions. Observers also highlighted threats from illicit foreign interference and disinformation.

These findings complicate attempts to tell a simple story in which one side's complaints cancel out the other's. An election can be well organised on polling day and still have unequal campaign conditions. External interference can be a serious problem while domestic authorities remain responsible for improving the rules.

The diaspora's role raises another political challenge. Moldovans abroad are part of the legal electorate. Their votes count towards the national result, not towards a separate result which can be discarded when it changes the winner.

Yet the government will have to address the concerns of people living inside the country whose experience of reform, wages or public services may differ from that of relatives overseas. Defending the legitimacy of diaspora voting does not remove the need to persuade those citizens.

The practical task for the European camp is therefore larger than celebrating a constitutional majority. It has to make the advantages of integration visible enough that the next contest does not depend on rescuing a result at the final stage of the count.

Money, influence and the evidence

Before polling, Moldovan law enforcement said it had uncovered a large vote-buying operation linked to Ilan Shor, the fugitive businessman living in Russia. Authorities described money transfers intended to influence both the referendum and the presidential vote.

Sandu accused criminal groups working with hostile foreign forces of trying to undermine the election. Moscow denied interference and challenged the allegations. Shor has denied wrongdoing. His activities and the extent of any wider coordination remain matters for evidence and legal proceedings.

On 25 October, police chief Viorel Cernăuțeanu said more than $24 million had been transferred from Russia during October in connection with the alleged network. That was an official investigative claim about financial flows, not a finding that every recipient cast a particular ballot.

The distinction is essential in a secret vote. Evidence of payment, instructions or registration in a network can establish an attempt to corrupt the process. It does not allow an observer to reconstruct every recipient's choice in the voting booth.

This creates two different questions for investigators. One concerns the organisation, financing and intended beneficiaries of the operation. The other concerns the extent to which it changed participation or outcomes. The first can sometimes be established more directly than the second.

The close referendum margin makes both questions urgent. It also makes exaggerated precision dangerous. A claim that an exact number of votes was bought should be supported by a clear account of how that number was derived, rather than inferred from the size of a contact list.

Police also rejected the suggestion that an anonymously circulated list of alleged participants demonstrated a split between law enforcement institutions. Prosecutors said they would examine the information against case materials. That response shows why leaked accusations should not be treated as an official verdict against everyone named.

The European Union's 22 October statement backed Moldova's European aspirations and condemned Russian interference. For Brussels, the issue is not only a single close vote. It is whether a candidate country can make decisions through institutions protected from covert financial pressure.

An effective response has to preserve due process as well as improve enforcement. If people are accused without an opportunity to challenge evidence, anti-interference measures can themselves become a source of mistrust. If credible financing schemes go unexamined, organisers learn that the political benefit outweighs the legal risk.

Neither requirement is a luxury before the runoff. Investigators need to disrupt unlawful activity quickly enough to protect the next ballot, while providing a public account that can withstand scrutiny after election day.

The same discipline applies to political attribution. Support from a network for a candidate does not, by itself, prove that the candidate directed it. Stoianoglo has denied ties to Shor. Any claim of coordination must be established through evidence of actual relationships and conduct.

The runoff asks a different question

The presidential first round gave Sandu a substantial lead over Stoianoglo but left a large pool of votes with other candidates. The second round removes those alternatives. Voters who preferred someone else must now choose between the finalists, abstain or cast an invalid ballot.

Adding all the losing candidates' totals to one finalist is not a forecast. Political endorsements can matter, but their supporters are not transferable possessions. Some are motivated by attitudes towards Sandu, others by local loyalties, policy or dissatisfaction with the entire political class.

Stoianoglo is trying to occupy a position broader than a simple rejection of Europe. In the late-October debate, he spoke of maintaining relations with Moldova's different development partners, including the EU, Ukraine, Romania, the United States, Russia and China.

He also proposed a joint declaration supporting European integration. Sandu challenged the sincerity of that position, pointing to his refusal to participate in the referendum. The exchange captures the campaign's central disagreement: whether he offers a different route towards Europe or a way to weaken the existing one.

The Socialist Party's backing gives Sandu material for her argument about geopolitical direction. It does not relieve her of the need to address domestic grievances. A voter can be unconvinced by Moscow's role and still dissatisfied with how the current authorities have governed.

For Stoianoglo, the difficulty runs the other way. Criticising the incumbent is easier than explaining how his proposed balance among foreign partners would work where their demands conflict. Relations with Russia and the EU cannot always be improved simultaneously through a general promise of neutrality or dialogue.

The presidency matters to foreign policy and the public direction of the state, but it does not govern alone. Parliament and the cabinet remain essential to legislation, budgets and the reforms required for accession. A presidential victory would alter the political balance without automatically deciding every institution's course.

That makes the contest both consequential and narrower than some campaign rhetoric implies. A Sandu victory would renew her mandate but leave the close referendum and dissatisfaction to address. A Stoianoglo victory would create a new political relationship with the existing government and the European process, rather than instantly replacing all of it.

The second-round campaign is also a test of participation. The first referendum showed how strongly the diaspora can affect the overall tally. Candidates need to mobilise citizens across different countries and time zones as well as at home.

There is a risk that disputes over turnout become disputes over which citizens are considered authentic members of the electorate. A successful campaign may exploit differences between communities; a president must then govern a country connected to all of them.

The safest political assumption for either finalist is that the first-round count is an opening position. The voters who decide the runoff are not obliged to repeat a party's preferred interpretation of 20 October.

Europe's promise meets Moldova's timetable

Ten days before the election, the European Commission announced a proposed €1.8 billion growth package for Moldova for 2025–2027. It linked greater financial support to reforms, investment and closer integration with the European economy.

The package identified potential improvements in transport links, including roads, rail connections and bridges over the Prut. These are the kinds of projects through which an abstract promise of European integration can become a practical change in how people travel, trade and reach work.

The announcement did not mean the entire amount had been transferred. The proposed facility and its implementation still required further steps, and funding would be tied to an agreed reform agenda. That difference between an announced envelope and completed projects is politically important.

Moldovan voters are being asked to support a process whose largest benefits may take time. Governments campaigning against them can offer immediate payments or promises without accepting the same institutional requirements. That is an asymmetry the EU cannot solve through larger headline figures alone.

It can respond through credible delivery. A completed connection, a transparent grant or a service that works is harder to dismiss than a distant accession date. But credible delivery also means explaining conditions and limits, so that unspent commitments are not presented as money already improving household finances.

The narrow referendum majority should encourage that practical approach. It provides permission to continue the European course. It does not show that every community trusts the government's account of its benefits.

The security environment adds pressure. Russia's war against neighbouring Ukraine makes Moldova's external orientation more consequential, while also giving domestic politicians a powerful language of threat. EU support needs to strengthen the country's institutions without treating every political disagreement as proof of disloyalty.

For Sandu, the immediate deadline is 3 November. She has a referendum result she can defend, but the presidential majority still has to be assembled. For Stoianoglo, the opportunity is to turn dissatisfaction into support while convincing voters that his foreign policy would protect their interests.

For Moldova's electoral authorities, the task is less rhetorical: administer another ballot, protect its integrity and publish a result citizens can examine. The first vote produced a yes to Europe by 10,564 votes. It left the question of who will lead the country along that path to a second Sunday.