More than half a million voters in Serbia’s electoral register were listed without a complete street name and house number, according to an investigation first published by Radio Free Europe/Radio Liberty (RFE/RL) on September 25. The finding has acquired renewed urgency ahead of the October 25 parliamentary election after election-monitoring group CRTA warned on October 22 that the voter register would be used without a completed audit. The figure describes incomplete address records; it does not establish that 500,000 voters are fictitious or that 500,000 votes will be cast fraudulently.
- A September investigation, revisited in the final days before voting
- What explains incomplete addresses — and what remains unverified?
- The authorities’ position and the unanswered questions
- What CRTA reported on October 22
- Why the distinction matters on October 25
- What voters and observers should watch next
- Sources and reporting note
The distinction matters. The question is not simply how many records contain an address anomaly, but which errors are explainable, whether independent reviewers can test the remaining cases, and whether a lack of transparency leaves opportunities for misuse.
A September investigation, revisited in the final days before voting
On September 25, 2026, RFE/RL journalists Jelena Janković and Nevena Bogdanović published an analysis of publicly available figures from Serbia’s Ministry of Public Administration and Local Self-Government. Their automated examination identified more than 500,000 voters recorded without a street name and house number. In many entries the address fields effectively read “no street, zero.” The investigation was not newly released on October 23; a fresh RFE/RL election overview published on October 23 has brought it back into the pre-election debate.
The original reporting found roughly 85,000 such entries in 13 border municipalities. Among the examples was Ivanje, in the municipality of Prijepolje, where records examined on September 21 put 1,523 voters under the same non-specific address. The settlement had 1,101 inhabitants in Serbia’s 2022 census, although the figures cannot be compared directly without knowing how many registered residents live abroad and how each record was classified. The national address register lists multiple named streets in the settlement.
RFE/RL also described apparently crowded addresses, including 247 registered voters at an address in Priboj associated with a family house, and hundreds registered at social-welfare centers. These are reasons to check the records, not sufficient evidence of unlawful registration. Serbian law can permit people without a conventional home address to register through a social-welfare center under specified conditions.
Sources: RFE/RL’s original September 25 investigation and its October 23 election overview.
What explains incomplete addresses — and what remains unverified?
CRTA researcher Vladimir Erceg told RFE/RL that one administrative explanation is the mismatch between the Interior Ministry’s residence database and the Republic Geodetic Authority’s address register. A newly assigned street name or house number may be recorded by one body before it appears in the other. Updating the voter register then depends on that chain of records being reconciled.
Erceg also described cases where voters previously listed without a house number were later moved to house number 1 as a group. CRTA’s checks in those examples found recording errors rather than proof of fictitious voters. Yet the same gaps complicate efforts to verify other entries and may conceal improper registrations if cross-checking cannot be completed.
The public information also has important limits. RFE/RL reported that apartment numbers are missing from more than 90% of Belgrade address entries examined, leaving the publicly available totals at building-entrance level rather than individual apartments. That reduces the usefulness of a citizen check intended to reveal how many people are registered in a household.
Incomplete address data, an unlawfully registered resident, a duplicate entry and an illegally cast ballot are four different claims. Each would require separate evidence. Public totals do not establish whether an individual actually resides at a location, voted at all, or voted more than once.
The authorities’ position and the unanswered questions
In the original RFE/RL reporting, the Ministry of Public Administration and Local Self-Government said residence and temporary-residence records fell within the Interior Ministry’s responsibility. The Interior Ministry had not answered RFE/RL’s detailed questions by publication. The governing Serbian Progressive Party rejected accusations that the voter register was being manipulated and described it as transparent and up to date. Those responses should be considered alongside, not substituted for, independent verification of the database.
Opposition representatives and independent monitors have sought broader access to underlying records to identify legitimate discrepancies, incorrect entries and any misuse. The existence of competing political claims does not settle the administrative facts: verifying the register requires documentary checks across the relevant state systems.
What CRTA reported on October 22
CRTA’s preliminary election-campaign report published October 22 said the register had been closed with 6,450,523 voters and without a completed audit. According to the monitoring organization, the audit commission did not receive the complete source-data access needed to verify the register effectively.
CRTA challenged a publicly cited 98% match figure from the government’s Office for IT and eGovernment, saying it was not approved by the commission and did not measure potential duplicate registrations or fictitious residences. The organization also said a review of a sample of voter-register extracts did not find voters improperly added or omitted, while identifying a different operational concern: more than 30,000 names were located outside the expected alphabetical order in polling-station extracts. It warned that polling boards should conduct additional checks before declaring an eligible voter absent from a list.
The October report therefore adds a second layer to the September address investigation. The earlier work identified incomplete public address fields; the later report raised a broader issue about the extent of institutional scrutiny possible before election day. Neither finding permits the conclusion that all anomalous records correspond to illegal votes.
Why the distinction matters on October 25
Questions about voter-register accuracy can have opposing practical consequences. Incorrect or unjustified entries can weaken confidence in the integrity of an election, while mistaken assumptions that unfamiliar or unconventional addresses are necessarily fraudulent can unfairly cast suspicion on legitimate voters, including people in informal settlements or receiving social assistance.
For election administrators and observers, the most useful test is traceable evidence: whether address discrepancies can be reconciled with legal residence records, whether corrections are documented, and whether polling-station lists allow eligible people to vote without creating opportunities for duplication.
Independent scrutiny will continue. The OSCE’s Office for Democratic Institutions and Human Rights (ODIHR) opened an election observation mission in Serbia on September 16. Its assessments and any subsequent audit findings should be read alongside official election data, not replaced by assumptions about the meaning of the 500,000 figure.
What voters and observers should watch next
As Serbia votes on Sunday, October 25, the outstanding questions are whether the Interior Ministry and other competent bodies provide a verifiable explanation for the incomplete addresses; whether election observers can obtain sufficient underlying data to distinguish administrative mistakes from suspected abuse; and whether polling-station teams consistently check full extracts before refusing an eligible voter.
Bottom line: RFE/RL documented a substantial transparency problem in September, and CRTA’s October 22 report says the register enters election day without a completed audit. These are meaningful findings about oversight. They are not evidence that half a million fraudulent votes exist, nor are they a prediction of the election outcome.
Follow Votendi’s Serbia election hub for election information, reporting and official results as they become available.
Sources and reporting note
- Radio Slobodna Evropa / RFE/RL, September 25, 2026 — original investigation and public-register analysis.
- RFE/RL, October 23, 2026 — updated election-period reporting.
- CRTA Election Observation Mission, October 22, 2026 — pre-election audit status and observation findings.
- CRTA, October 11, 2026 — background on record access and register verification.
- OSCE/ODIHR, September 16, 2026 — opening of Serbia election observation mission.
Reporting cut-off: October 23, 2026. The 500,000-plus estimate is attributed to RFE/RL’s September analysis of public data, not an independently conducted Votendi audit. Claims of manipulation are reported as allegations; no inferred fraudulent-vote total or electoral effect is asserted. Featured image is a previously published Votendi conceptual illustration of election observers in Belgrade, not a photograph of the voter register or of a documented incident.
