In the run-up to the elections on 25 October, the voter register will remain a source of doubt and public distrust even after it is closed on 9 October. Public data show that the number of voters, after years of decline, has begun to rise again ahead of the elections, most of all in Belgrade. A record number of voters have registered to vote at a temporary residence in Serbia and abroad. The first comprehensive analytical findings by members of the Commission for the audit of the voter register show that the register contains a large number of records that need to be checked, but there is neither the time nor the conditions to do so before the elections. Nor was the proposal adopted that the Commission should receive, before the elections, the extracts from the voter register used for voting so that it could check them.
Citizens have become the first line of control over the register at the level of their own addresses, but the path from a problem spotted to a correction made remains unclear to the great majority. Voting from abroad drew the most attention: an absolute record, almost three times as many registrations, was received by a network of polling stations that was expanded, through the intervention of the Ministry of Foreign Affairs, by only 15%, and some 22,000 of those who applied were not covered by the decision on polling stations.
Citizens have become the first line of control over the register at the level of their own addresses, but the path from a problem spotted to a correction made remains unclear to the great majority. Voting from abroad drew the most attention: an absolute record, almost three times as many registrations, was received by a network of polling stations that was expanded, through the intervention of the Ministry of Foreign Affairs, by only 15%, and some 22,000 of those who applied were not covered by the decision on polling stations.
Changes in the voter register
Public snapshots show that the number of voters, after years of decline, has begun to rise again in the months before the elections, and most of all in Belgrade. These data do not show that the changes are unlawful, but they call for an explanation, and there is still no answer.
According to the public snapshots of the Ministry of Public Administration and Local Self-Government, the number of voters outside Kosovo and Metohija fell from 6,395,114 in October 2021 to 6,326,661 in February 2026, and then rose by 6,994 by 28 September. In the same period in 2025 the number fell by 1,682. Ministry of Public Administration and Local Self-Government, public snapshots of the number of voters by local self-government unit, October 2021 - 28 September 2026, processed by CRTA. The snapshots were previously published quarterly, then monthly, and are today published weekly. Part of the growth was to be expected, since after the elections are called persons who reach the age of majority by election day are entered in advance. Law on the Unified Voter Register, “Official Gazette of the RS”, Nos. 104/2009, 99/2011, 44/2024 and 96/2025, Article 4(1). Even so, of a net increase of around 7,000 voters, more than 6,400 relates to Belgrade municipalities, which make up about a quarter of the register. The furthest above last year's growth is Savski venac (+547, against -188 a year earlier). The rule that a voter who changes permanent residence is entered in the new municipality only after six months applies solely to voters who are already in the register and are moving between two municipalities. Persons being entered for the first time, following a first registration of permanent residence or the acquisition of citizenship, may be entered without restriction right up to the closing of the register. Law on the Unified Voter Register, Articles 4(2) and 6(2).
The public data on changes by legal ground, published since 15 February 2026, single out three phenomena: Ministry of Public Administration and Local Self-Government, public data on the number of changes to the voter register by legal ground (Article 14 of the Law on the Unified Voter Register), published since 15 February 2026; for earlier years, the Ministry's monthly data; processed by CRTA.
According to the public snapshots of the Ministry of Public Administration and Local Self-Government, the number of voters outside Kosovo and Metohija fell from 6,395,114 in October 2021 to 6,326,661 in February 2026, and then rose by 6,994 by 28 September. In the same period in 2025 the number fell by 1,682. Ministry of Public Administration and Local Self-Government, public snapshots of the number of voters by local self-government unit, October 2021 - 28 September 2026, processed by CRTA. The snapshots were previously published quarterly, then monthly, and are today published weekly. Part of the growth was to be expected, since after the elections are called persons who reach the age of majority by election day are entered in advance. Law on the Unified Voter Register, “Official Gazette of the RS”, Nos. 104/2009, 99/2011, 44/2024 and 96/2025, Article 4(1). Even so, of a net increase of around 7,000 voters, more than 6,400 relates to Belgrade municipalities, which make up about a quarter of the register. The furthest above last year's growth is Savski venac (+547, against -188 a year earlier). The rule that a voter who changes permanent residence is entered in the new municipality only after six months applies solely to voters who are already in the register and are moving between two municipalities. Persons being entered for the first time, following a first registration of permanent residence or the acquisition of citizenship, may be entered without restriction right up to the closing of the register. Law on the Unified Voter Register, Articles 4(2) and 6(2).
The public data on changes by legal ground, published since 15 February 2026, single out three phenomena: Ministry of Public Administration and Local Self-Government, public data on the number of changes to the voter register by legal ground (Article 14 of the Law on the Unified Voter Register), published since 15 February 2026; for earlier years, the Ministry's monthly data; processed by CRTA.
The public data on the number of voters per address have also made the first independent analyses possible. According to an analysis by Radio Slobodna Evropa, more than 500,000 are entered at addresses with no street name and house number,
Radio Slobodna Evropa, 25 September 2026, accessed 6 October 2026, https://link.crta.rs/13w.
while, on the basis of access to one of the services that aggregate the Ministry's data, flat numbers are recorded for only around 2.5% of Belgrade addresses.
Danas, 12 September 2026, accessed 6 October 2026, https://link.crta.rs/13x.
A citizens' tool for checking by address singles out addresses with an unusually large number of voters and addresses of non-residential and public buildings,
bs.dignidupe.org, an unofficial tool based on data from the geoSrbija portal and upit.birackispisak.gov.rs, accessed 6 October 2026.
but without a link to the residence records it, too, cannot establish whether the voters at an address actually live there. Only changes visible from public data are set out here.
Status of the audit and the work of the Commission
Access to data
Despite three months of cooperation with the Office for IT and eGovernment, the scope of access to data has remained at the same level since August. There has still been no movement towards adequate access to the data from the source records to which the Commission is entitled. The Commission has broad access to the Unified Voter Register (UVR) itself, but it can check the source records only case by case.
CRTA Election Observation Mission, Campaigning without the paperwork, link.crta.rs/131 http://link.crta.rs/131
During the observation period, the Commission ordered that the cross-checking of the UVR against the source records, which is carried out by the Office for IT and eGovernment, should begin and that, after the pilot test in Sevojno, it should be extended to the whole of Serbia. The test is under way and the first results are expected in mid-October. In any event, the findings of that check will hardly affect the register on the basis of which voting will take place on 25 October.
First notes on the state of the data
On 6 October, members of the Commission presented four official notes: on access to data, on the description of the data in the UVR, on multiple entries, and on checking the accuracy of the data on samples in ten municipalities.
The Commission, Official note on the findings from analytical unit 3: Description of the data in the Unified Voter Register,; Official note on the findings from analytical unit 3: Multiple entries in the Unified Voter Register; Official note on the findings from analytical unit 5: Checking the accuracy of the data on the basis of a sample in 10 local self-government units (5 October 2026).
The finding on six municipalities in Kosovo and Metohija (Glogovac, Dečani, Đakovica, Kačanik, Suva Reka and Štimlje) deserves particular attention. A total of 248,042 voters from these municipalities are entered in the UVR, but since 2012 these municipalities have not been covered by elections or by the publicly available data on the register, and the legal ground for this is not known to the Commission. Two total figures for the number of voters are therefore cited in public: around 6.45 million, as shown by the public data, and around 6.7 million, which is the number of records in the database. On 28 September, the president of the REC stated that “around 6,440,000 voters” were entered in the register, while a member of the Commission, Ognjen Rašuo, stated a day later that there were 6,696,433, including around 250,000 voters in six municipalities in Kosovo and Metohija. “Dimitrijević for RTS: Everything is going according to plan, by 25 October everything will be ready for the elections”, RTS, 28 September 2026, accessed 6 October 2026, https://link.crta.rs/13y. Milica Božinović, “How the number of people in the voter register changes on a daily basis and how many voters are currently entered”, Nova.rs, 29 September 2026, accessed 6 October 2026. The data on these voters are, by all appearances, not being updated.
The other findings point to questions that are yet to be resolved:
The finding on six municipalities in Kosovo and Metohija (Glogovac, Dečani, Đakovica, Kačanik, Suva Reka and Štimlje) deserves particular attention. A total of 248,042 voters from these municipalities are entered in the UVR, but since 2012 these municipalities have not been covered by elections or by the publicly available data on the register, and the legal ground for this is not known to the Commission. Two total figures for the number of voters are therefore cited in public: around 6.45 million, as shown by the public data, and around 6.7 million, which is the number of records in the database. On 28 September, the president of the REC stated that “around 6,440,000 voters” were entered in the register, while a member of the Commission, Ognjen Rašuo, stated a day later that there were 6,696,433, including around 250,000 voters in six municipalities in Kosovo and Metohija. “Dimitrijević for RTS: Everything is going according to plan, by 25 October everything will be ready for the elections”, RTS, 28 September 2026, accessed 6 October 2026, https://link.crta.rs/13y. Milica Božinović, “How the number of people in the voter register changes on a daily basis and how many voters are currently entered”, Nova.rs, 29 September 2026, accessed 6 October 2026. The data on these voters are, by all appearances, not being updated.
The other findings point to questions that are yet to be resolved:
Extracts from the voter register remain unchecked
After the register is closed, the Ministry of Public Administration and Local Self-Government has 48 hours to prepare and certify printed extracts for each polling station and submit them to the Republic Electoral Commission (REC).
Law on the Unified Voter Register, Article 19.
An extract is the list of voters for a single polling station, in which the voter enters a signature when voting. At the Commission's session on 6 October, a discussion was held on a proposal that the extracts, in the form prepared for printing, should also be submitted to the Commission's secure server. Those who put the proposal forward pointed out that the Commission has no insight into the distribution of voters across polling stations, and that the audit methodology, which the Commission adopted unanimously on 3 April, provides for a check that the extracts are identical to the register in the system.
The Commission, Act on the procedure for the audit of the Unified Voter Register (3 April 2026), procedure 2.1.3.
The members appointed on the proposal of the ruling majority opposed this, arguing that the extracts fall within the competence of the REC and of the Ministry and that submitting them would put pressure on the deadlines and the logistics of printing, and as a compromise they proposed that the Commission review a sample of the extracts while these are at the REC.
CRTA, Election Observation Mission, observations from the session of the Commission, 6 October 2026.
Five members voted for the proposal (two appointed on the proposal of associations and three on the proposal of the opposition), four members from the ranks of the majority were against, and one abstained. The proposal was not adopted, because a decision of the Commission requires at least seven votes, including at least two members from the ranks of the majority.
Law on the Unified Voter Register, Article 22lj. The Commission is made up of ten members: five appointed on the proposal of the parliamentary groups of the ruling majority (SNS, SPS, PUPS, SDPS and SVM), three on the proposal of opposition parliamentary groups (ZLF, SSP and NPS) and two on the proposal of associations (CeSID and CRTA).
A review of a sample does not make it possible to check all the extracts, so the last link between the database and the polling station remains without independent verification before the elections, even though the Commission adopted the methodology containing that check unanimously.
Members of the Commission announced that they would present notes on comprehensive analyses of movements in the voter register, and, after the register is closed, an assessment of the changes in the period before the elections.
Members of the Commission announced that they would present notes on comprehensive analyses of movements in the voter register, and, after the register is closed, an assessment of the changes in the period before the elections.
From the closing of the register to the elections42
9 October: the Ministry of Public Administration and Local Self-Government closes the register and establishes the total number of voters, which the REC publishes in the “Official Gazette”.
by 11 October: the Ministry submits the certified extracts to the REC, including the extracts for voters from Kosovo and Metohija, who, as in 2023, will vote in Vranje, Raška, Kuršumlija and Tutin.
10-21 October: changes to the register only by a ruling of the Ministry, after which the REC publishes the final number of voters.
28 October: the statutory deadline for the Commission's first report to the National Assembly.
42 Law on the Unified Voter Register, Articles 17-20; Law on Amendments to the Law on the Unified Voter Register, “Official Gazette of the RS”, No. 96/2025, Article 8; Republic Electoral Commission, Decision of 5 October 2026 determining the manner in which voters from the territory of the Autonomous Province of Kosovo and Metohija will exercise their electoral right (5 October 2026).
Citizen reports concerning the voter register
Since the number of voters per address became publicly available, citizens have increasingly been checking their own buildings and houses themselves. The reports do not show how many unlawful entries there are in the register, but they do show that citizens can spot a problem while having almost no way of resolving it.
By 6 October, CRTA had received 341 messages from citizens, making up 190 cases for checking the voter register, almost twice as many as the previous report covered. CRTA, records of citizen reports on the voter register, March - 6 October 2026, including reports that did not arrive by email. The previous report covered 97 citizens who contacted CRTA between March and the beginning of September 2026. Some of the reports were forwarded to the Commission. A third of the cases concern incomplete and inconsistent address data: addresses without a house number, buildings without flat and entrance numbers, old and new street names. A further fifth concern the difference between the number of voters at an address and the number of people who actually live there. Next come voters entered in non-residential, empty or demolished buildings, suspicions of fictitious permanent residence, and deceased persons who are, according to the accounts, still in the register. Some of the suspicions turn out on checking to be unfounded or to be the consequence of poor address data, for example when a building with several entrances is recorded as a single address. Even so, when voters whom none of the residents know have been registered in the same building for years, the problem does not lie in the way the data are displayed but in the records themselves.
The path from a problem spotted to a correction made is unclear. One owner of a flat in Belgrade obtained written confirmation from the police that no one was registered at her address, while the public display on the same day showed 14 voters, and two months later it still showed the same. A residents' association in Zemun that asked for a joint check of the persons registered there was told by the police that each owner must come in person, with proof of ownership and a paid fee. Even so, when residents know what they are looking for, they can achieve a result. In two of the communications, addresses with manipulative entries were identified in conversation with citizens. At one, the residents' association managed to force the owner to de-register more than 30 persons, and at the other, according to the residents' accounts, they are organising to do the same. Such outcomes depend on the persistence of citizens rather than on systemic checking, and it is not known whether anyone has been held to account for registering the persons who were subsequently de-registered.
By 6 October, CRTA had received 341 messages from citizens, making up 190 cases for checking the voter register, almost twice as many as the previous report covered. CRTA, records of citizen reports on the voter register, March - 6 October 2026, including reports that did not arrive by email. The previous report covered 97 citizens who contacted CRTA between March and the beginning of September 2026. Some of the reports were forwarded to the Commission. A third of the cases concern incomplete and inconsistent address data: addresses without a house number, buildings without flat and entrance numbers, old and new street names. A further fifth concern the difference between the number of voters at an address and the number of people who actually live there. Next come voters entered in non-residential, empty or demolished buildings, suspicions of fictitious permanent residence, and deceased persons who are, according to the accounts, still in the register. Some of the suspicions turn out on checking to be unfounded or to be the consequence of poor address data, for example when a building with several entrances is recorded as a single address. Even so, when voters whom none of the residents know have been registered in the same building for years, the problem does not lie in the way the data are displayed but in the records themselves.
The path from a problem spotted to a correction made is unclear. One owner of a flat in Belgrade obtained written confirmation from the police that no one was registered at her address, while the public display on the same day showed 14 voters, and two months later it still showed the same. A residents' association in Zemun that asked for a joint check of the persons registered there was told by the police that each owner must come in person, with proof of ownership and a paid fee. Even so, when residents know what they are looking for, they can achieve a result. In two of the communications, addresses with manipulative entries were identified in conversation with citizens. At one, the residents' association managed to force the owner to de-register more than 30 persons, and at the other, according to the residents' accounts, they are organising to do the same. Such outcomes depend on the persistence of citizens rather than on systemic checking, and it is not known whether anyone has been held to account for registering the persons who were subsequently de-registered.
The voter register in the media
The voter register was one of the more visible election topics, but the reporting remained event-driven and polarised, while the causes of the problems and the work of the Commission as an institution went almost unexplained.
CRTA's review of publicly available online reporting on the voter register, 9 September - 6 October 2026, a purposive sample of 48 items from the public service broadcaster and from pro-government, critical, regional and international outlets. Television programming was not monitored. CRTA was a source of information in some of the items.
Attention grew in waves: from the statement by the Minister of Public Administration and Local Self-Government that nothing had come from the Commission to indicate any specific irregularity, Insajder, 9 September 2026, https://link.crta.rs/143. through data from opposition members of the Commission on municipalities with more voters than residents and local stories about “phantom voters”, Danas, 22 September 2026, https://link.crta.rs/144; N1, 30 September 2026, https://link.crta.rs/145. to voting from abroad, which dominated the reporting at the beginning of October. Outlets close to the authorities described doubts about the register as “conspiracy theories”. When, on 23 September, several pro-government outlets used the same frame in reporting a statement by a CRTA representative that the current volume of new entries cannot have a decisive effect on the election result, presenting it as proof that the doubts were unfounded, CRTA's assessments from the same period about serious weaknesses in the register were left out. Večernje novosti, 23 September 2026, https://link.crta.rs/13z; B92 https://link.crta.rs/140; Kurir, 23 September 2026, https://link.crta.rs/141; Informer, 23 September 2026, https://link.crta.rs/142. The public service broadcaster reported procedurally, and critical outlets also published analyses based on public data, but the headlines most often erased the difference between disorderly records and manipulation. For example: RTS, 28 September 2026; Vreme, 26 September 2026, https://link.crta.rs/147; Nova.rs, 12 September 2026, https://link.crta.rs/146.
Attention grew in waves: from the statement by the Minister of Public Administration and Local Self-Government that nothing had come from the Commission to indicate any specific irregularity, Insajder, 9 September 2026, https://link.crta.rs/143. through data from opposition members of the Commission on municipalities with more voters than residents and local stories about “phantom voters”, Danas, 22 September 2026, https://link.crta.rs/144; N1, 30 September 2026, https://link.crta.rs/145. to voting from abroad, which dominated the reporting at the beginning of October. Outlets close to the authorities described doubts about the register as “conspiracy theories”. When, on 23 September, several pro-government outlets used the same frame in reporting a statement by a CRTA representative that the current volume of new entries cannot have a decisive effect on the election result, presenting it as proof that the doubts were unfounded, CRTA's assessments from the same period about serious weaknesses in the register were left out. Večernje novosti, 23 September 2026, https://link.crta.rs/13z; B92 https://link.crta.rs/140; Kurir, 23 September 2026, https://link.crta.rs/141; Informer, 23 September 2026, https://link.crta.rs/142. The public service broadcaster reported procedurally, and critical outlets also published analyses based on public data, but the headlines most often erased the difference between disorderly records and manipulation. For example: RTS, 28 September 2026; Vreme, 26 September 2026, https://link.crta.rs/147; Nova.rs, 12 September 2026, https://link.crta.rs/146.
