Update
How CRTA observes elections: inside the methodology and stru...
Standards, methodology and structure of the mission CRTA deploys for the 2026 elections
Between the published vote totals and the published seat totals there is one division, and it is the step nobody shows. Each list's votes are divided by one, two, three and so on, and the highest quotients that come out take the seats. The table below does that in the open, one seat at a time, and says why each seat went where it did. For a full election result, with real lists, coalitions and every cycle since 2008, the seat calculator runs the same arithmetic at scale.
The divisions
Scroll the table sideways to see every list.
0 of 15 seats allocated
Seats to fill
Where that leaves everyone
| List | Votes | Vote share | Seats | Seat share | Difference |
|---|
The same arithmetic, at full size
The National Assembly has 250 seats, so the same division runs until 250 quotients have been used up. Nothing about the method changes; only the number of steps does. At the 2023 parliamentary elections the line fell here.
The last seats awarded, and the first quotients that missed
| Seat 247 | Ivica Dačić - Prime Minister of Serbia | 249,916 ÷ 18 | 13,884.2 |
| Seat 248 | Serbia Against Violence | 902,450 ÷ 65 | 13,883.8 |
| Seat 249 | Aleksandar Vučić - Serbia Must Not Stop | 1,783,701 ÷ 129 | 13,827.1 |
| Seat 250 | We - Voice of the People | 178,830 ÷ 13 | 13,756.2 |
| Missed | Aleksandar Vučić - Serbia Must Not Stop | 1,783,701 ÷ 130 | 13,720.8 |
| Missed | Nada for Serbia | 191,431 ÷ 14 | 13,673.6 |
| Missed | Serbia Against Violence | 902,450 ÷ 66 | 13,673.5 |
The 250th seat went to the list of Branimir Nestorović on a quotient of 13,756.2, and the next quotient in line, the 130th division of the governing list, came out at 13,720.8. Had the Nestorović list won 460 votes fewer, that seat would have moved across. Its thirteenth seat rested on a margin of four hundred and sixty voters out of 3.7 million, and on nothing that list itself did on the day.
The other side of the same arithmetic is what never entered it at all: 260,905 votes, 7.0 per cent of the valid ballots, went to lists that stayed under the three per cent threshold. Those votes were not transferred anywhere. They were simply left out of the divisions, which is why the lists that did pass ended up with a larger share of the seats than of the votes.
A vote total is divided by one, two, three and so on up to the number of seats being allocated. Lists below the three per cent threshold take no part in the division. Lists of national minorities are exempt from the threshold and have their quotients increased by 35 per cent. The figures for 2023 are the official results of the Republic Electoral Commission. Where two quotients come out exactly equal, the demonstration above gives the seat to the list higher up the table; with real vote totals an exact tie is vanishingly unlikely.
Once the seats start moving, the usual explanation of the Serbian result gets harder to defend. At the 2023 parliamentary elections the governing list won 48.1 per cent of the valid ballots and 51.6 per cent of the seats, and that gap of three and a half percentage points is routinely attributed to d'Hondt. Measured against the votes that actually entered the allocation, however, it took 51.7 per cent and received 51.6 per cent of the seats - marginally less than its arithmetical share. The gap came almost entirely from the three per cent threshold, not from the method.
The second tool is about that threshold. Votes below it are not passed on to anyone, so the question worth asking is who ends up with the seats they would have won. Move the line, or carry a single list across it, and the table answers in seats and in the number of votes the move took. One caution holds for all of it: the comparison keeps voter behaviour constant. It asks what the same ballots would have produced under a different rule, not how citizens would have voted had the rule been different.
The gap came almost entirely from the three per cent threshold, not from the method.
The second tool is about that threshold. Votes below it are not passed on to anyone, so the question worth asking is who ends up with the seats they would have won. Move the line, or carry a single list across it, and the table answers in seats and in the number of votes the move took. One caution holds for all of it: the comparison keeps voter behaviour constant. It asks what the same ballots would have produced under a different rule, not how citizens would have voted had the rule been different.
Seats, 2023
Scroll the table sideways to see every column.
| List | Votes | Share | Seats | Against 2023 |
|---|
Lists of national minorities are the exception, and the way they are treated is often described incorrectly. They are exempt from the three per cent threshold, but they do not enter through a natural quota, as they did before 2020. Since the amendments of that year their quotients are increased by 35 per cent when seats are allocated, which is the more favourable of the two rules. Exemption is not a guarantee either: at the 2023 elections one minority list took part in the allocation and still finished without a seat.
Both rules come down to a number of votes, and that number follows from the turnout. The last tool works it out: how many voters are on the register, how many of them vote, how many ballots are spoilt, and what a list therefore has to reach - three per cent of the valid votes to clear the threshold, or about a quarter less than the price of a seat if it is a list of a national minority. The threshold is exact, because the law measures it against the valid votes. The two prices are ceilings: everything that stays below the threshold leaves the division and makes the remaining seats cheaper, which in 2023 brought the price of a seat down by about a thousand votes.
Both rules come down to a number of votes, and that number follows from the turnout. The last tool works it out: how many voters are on the register, how many of them vote, how many ballots are spoilt, and what a list therefore has to reach - three per cent of the valid votes to clear the threshold, or about a quarter less than the price of a seat if it is a list of a national minority. The threshold is exact, because the law measures it against the valid votes. The two prices are ceilings: everything that stays below the threshold leaves the division and makes the remaining seats cheaper, which in 2023 brought the price of a seat down by about a thousand votes.
How many votes it takes
Voters on the registerUnified Voter Register, September 2026
TurnoutShare of them who vote
%
Invalid ballotsUsually between 2 and 2.5 per cent
%
To clear the threshold
113,010
Three per cent of the valid votes. Below this a list wins nothing, whatever its result.
A minority list’s first seat
11,161
Lists of national minorities do not have to clear the threshold. Their quotients are raised by 35 per cent instead, so their first seat costs about a quarter less than anyone else’s.
Votes cast: 3,863,582
Valid votes: 3,766,993
One seat, at most: 15,068
None of this is a defect in the count. The highest-quotient system is used across Europe, its properties are known, and the threshold is a deliberate choice written into the Law on parliamentary elections rather than an accident of the arithmetic. What the arithmetic cannot do is be sounder than the numbers entering it, and those are what CRTA documents - the accuracy of the register, the conditions in which signatures are collected, the composition of polling boards and the counting itself.
