Spain General election29 Nov 2026in 8 days PP + Vox majority 99% 2023 investiture bloc 1% Calendar Track record Research
Research · Spain
EN ES

From votes to seats: D'Hondt in 52 provinces

Abstract. We implement the LOREG's seat rules: 350 seats, one each for Ceuta and Melilla, the rest apportioned to provinces with a minimum of two, and D'Hondt within each province among lists with 3% of the valid vote. Given the official votes, the calculation reproduces all 350 seats of 2023, and every election since 2004. D'Hondt in small provinces gives the largest parties a bonus and leaves middling national parties short. National polls reach the provinces by proportional swing from 2023, with Sumar's 2023 vote split between Sumar and Podemos; this projection has not been backtested.

Polling Forecast · updated 9 November 2026 · 11 min read · Spain, Congreso, seat allocation, LOREG, D'Hondt, 3% threshold, provinces, threshold of exclusion, proportional swing, Sumar, Podemos

Summary

This paper describes how Polling Forecast turns national vote shares into a Congreso de los Diputados. The calculation has two halves. The first is the law: 350 seats shared among 52 provinces, D'Hondt within each province among lists with 3% of the valid vote. Our implementation of that law, given the official 2023 votes, reproduces the official result in every province, and a separate check reproduces all 350 seats at each general election since 2004. The second half is a projection, from today's national polls to vote shares in each province. That half is ours: a proportional swing from 2023, with rules for the parties that have split or appeared since then. It has not been tested against a past election.

The law is where Spanish seats are won and lost. Most provinces elect five deputies or fewer, and in those the D'Hondt count gives the largest parties more than their share and small national parties little or nothing. A national party on 6% can finish with a handful of seats; two parties that split the same vote can win fewer seats between them than one list would. Both effects are sharp, and both are modelled exactly.

What has changed since this paper was written

Updated 9 November 2026. The rest of the paper is as first published, except where a sentence below says it was corrected.

  • The map (November). The map can be coloured by any census figure; each census figure also has its own page (census maps). It can also replay a past polling miss ("The polls miss again") or take a vote you set.

Fifty-two contests

A Spanish general election is 52 separate contests. Each voter chooses one closed list in their province; the seats of that province go to the lists in proportion to their votes; and the party totals in the Congreso are the sums of the 52 results. Nothing is shared nationally. There is no compensation for a party whose votes are spread thinly, and no national threshold.

The rules are in articles 162 and 163 of the LOREG (Ley Orgánica 5/1985 del Régimen Electoral General).

Seats per province

Ceuta and Melilla elect one deputy each. Each of the other 50 provinces starts with two seats. The remaining 248 are shared in proportion to population: the national population is divided by 248, each province receives the whole number of times its population contains that quota, and the seats still left go to the provinces with the largest remainders. The apportionment is fixed in the decree that calls each election. For 2023 it was Real Decreto 400/2023, which gave Madrid 37 seats, Barcelona 32, Valencia 16, and Soria 2.

The two-seat minimum weights the chamber towards the thinly populated provinces of the interior. It also makes most provinces small:

Seats in the provinceProvincesSeatsShare of the Congreso
1 (Ceuta, Melilla)221%
2–39267%
4–5188023%
6–91611232%
10–1956117%
20 or more26920%
Live Provinces by number of seats (the 2023 apportionment) Computed from the database, polls to 19 November 2026.

29 provinces elect five deputies or fewer; together they elect about three in ten of the Congreso's members.

D'Hondt

Within a province the seats go by the D'Hondt method. Each list's votes \(v_p\) are divided by 1, 2, 3 and so on, and the seats go to the largest quotients. Equivalently, seats are handed out one at a time to the list with the largest

\[ \frac{v_p}{s_p + 1}, \tag{1} \]

where \(s_p\) is the number of seats it has so far. When two quotients are equal the seat goes to the list with more votes (LOREG art. 163.1.d), and our code breaks ties the same way. An exact tie in votes, which the law settles by lot, has never arisen and is not modelled.

The threshold

A list needs 3% of the valid vote in the province to take part, and blank votes count as valid (LOREG arts. 96.5 and 163.1.a). Ceuta and Melilla elect one deputy each, and the list with the most votes wins, which is what D'Hondt does with one seat.

A small province at work: Huesca, 2023

Huesca elects three deputies. In 2023 the shares of the valid vote were PP 38.2%, PSOE 33.6%, Vox 12.6%, Sumar 11.5%, others 3.1% and blank 1.0%. The quotients were:

List ÷1 ÷2 ÷3
PP 38.2 19.1 12.7
PSOE 33.6 16.8 11.2
Vox 12.6 6.3 4.2
Sumar 11.5 5.7 3.8

Table: D'Hondt quotients in Huesca, 2023, as shares of the valid vote. The three largest, in bold, take the seats.

The PP took two seats and the PSOE one. Vox and Sumar, with 24% of the vote between them, took none. The third seat went to the PP's second quotient, 19.1, ahead of the PSOE's 16.8: the PSOE would have needed 38.2% for a second seat, 4.6 points more than it had. For Vox or Sumar to take a seat, either would have needed more than 19.1%, about half as much again as Vox had.

Two properties of D'Hondt follow from the table, and they drive the Spanish results.

The threshold of exclusion. A list with more than \(1/(M+1)\) of the vote in an \(M\)-seat province is sure of a seat whatever the others do. In Huesca that is 25%; in a two-seat province, 33%. Below it, a seat depends on how the rest of the vote is split. The lowest share that can win a seat in the most favourable split, the threshold of representation, is \(1/(M + P - 1)\) for \(P\) lists (Rae, Hanby and Loosemore, 1971). Both fall as provinces grow, which is why the legal 3% threshold matters so little: it is far below what any list needs in a small province. With four or five serious lists, 3% can only bind where there are more than about 30 seats, which means Madrid and Barcelona.

The bonus for the largest lists. Among the usual divisor methods, D'Hondt is the one most favourable to large parties (Balinski and Young, 1982). In a small province the largest list often takes a seat with a quotient built on a remainder that would have been too small for anyone else. Added over 52 provinces, the effect is large:

Party2023 votes2023 seatsVotes todaySeats today
PP33.139.130.837.1
PSOE31.734.627.632.9
Vox12.49.419.019.1
Sumar12.38.96.12.0
ERC1.92.02.53.1
SALF––2.00.0
Podemos––1.50.0
EH Bildu1.41.71.52.3
Junts1.62.01.31.4
BNG0.60.31.00.6
PNV1.11.40.81.1
CC0.50.30.40.3
UPN0.20.30.00.0
Live Votes and seats: 2023 and today's projection (share of the valid vote and share of the 350 seats, %) Computed from the database, polls to 19 November 2026.

In 2023 the PP turned 33.1% of the vote into 39.1% of the seats and the PSOE 31.7% into 34.6%; Vox turned 12.4% into 9.4% and Sumar 12.3% into 8.9%. Parties whose vote is concentrated in a few provinces (ERC, Junts, EH Bildu, the PNV) win their share or a little more, because in their own provinces they are large parties; the BNG and Coalición Canaria, smaller even in their own provinces, win less. The main losers are national parties of middling size, whose vote is spread evenly over provinces where it is too small to win anything.

The check: 2023, and every election since 2004

A seat calculation that cannot reproduce the last election cannot be trusted with the next. The site's code (SpainForecast.DHondt) was given the official 2023 votes of every list in every province, from the Junta Electoral Central's summary in the BOE, with the 2023 number of seats per province:

PartyOur calculationOfficial
PP137137
PSOE121121
Vox3333
Sumar3131
ERC77
Junts77
EH Bildu66
PNV55
BNG11
UPN11
CC11
Total350350
Provinces where every seat matches5252
Live The seat law run on the official 2023 votes, against the official result Computed from the database, polls to 19 November 2026.

Every province matches: the same parties win the same number of seats in all 52. A separate script in our data pipeline, written independently, ran the same rule on every general election from 2004 to 2023, keeping every candidacy as published, and reproduced the official seats in all 52 provinces at all eight (The Spanish data: results, maps, statistics, deputies and polls).

This check covers the law, not the forecast. It says that given the right votes we produce the right Congreso. It says nothing about whether our projected votes are right, which is the subject of the rest of this paper.

From national polls to provinces

The seat law needs vote shares in each province. Polls give national shares for the PP, the PSOE, Vox, Sumar, Podemos, SALF, ERC, Junts, EH Bildu, the PNV, the BNG, Coalición Canaria, UPN and "others" (see Forecasting the Congreso from imperfect polls: an overview for the average). The projection fills the gap by proportional swing from 2023.

Ratios

Let \(S_p\) be today's national share for party \(p\) and \(B_p\) its 2023 national share of the valid vote. Others is whatever the named parties leave, with a floor: \(S_{\text{oth}} = \max(0.5,\ 100 - \sum_p S_p)\). Blank votes sit inside others, as they do in the pollsters' published estimates. Each party's swing ratio is

\[ r_p = \begin{cases} S_p / B_p & B_p > 0.05,\\ 1 & \text{otherwise,}\end{cases} \qquad r_{\text{oth}} = \frac{S_{\text{oth}}}{B_{\text{oth}} + B_{\text{blank}}}. \tag{2} \]

Provinces

Let \(b_{k,p}\) be party \(p\)'s 2023 share of the valid vote in province \(k\), with blank votes kept as a separate entry that moves with others. The projected share is

\[ z_{k,p} = 100\,\frac{b_{k,p}\, r_p}{\sum_q b_{k,q}\, r_q}, \tag{3} \]

and the seats in province \(k\) are D'Hondt on the \(z_{k,p}\) among the named parties with \(z_{k,p} \geq 3\). "Others" and blank votes take part in the total but never win seats: others is a residual, not a list.

This is the character of proportional swing. A party gaining a third of its vote nationally gains a third in every province, so it gains most points where it was strongest; a party losing ground loses most where it had most. No share falls below zero, which matters in a system where the regional parties stand in a handful of provinces and are absent everywhere else. Under uniform swing (the same number of points added everywhere), a national party losing five points would go below zero wherever it had less than five.

Parties that split: Sumar and Podemos

Sumar stood in 2023 as a single list in all 52 provinces, and Podemos was part of it. Since Podemos left Sumar's parliamentary group in December 2023, pollsters have published them separately. Neither has a 2023 result of its own, so the projection treats them as the lineage they came from:

  1. the swing ratio for the lineage is their combined share today over Sumar's 2023 share, \(r_{\text{left}} = (S_{\text{Sumar}} + S_{\text{Podemos}}) / B_{\text{Sumar}}\);
  2. in each province, Sumar's 2023 share is scaled by that ratio, as in equation (3);
  3. the result is split between the two in their national ratio today: Podemos takes the fraction \(S_{\text{Podemos}} / (S_{\text{Sumar}} + S_{\text{Podemos}})\) and Sumar the rest.

Each is then a separate list in the D'Hondt count, with its own 3% threshold. The split is where the forecast feels the full weight of D'Hondt. Each part of the left needs to clear the threshold of exclusion on its own, and in most small provinces neither does:

If they standSumarPodemosTogetherPSOEPPVox
apart (our projection)70711513067
as one list1101111412767
Live Sumar and Podemos: today's votes as two lists and as one Computed from the database, polls to 19 November 2026.

On today's polls, standing apart costs Sumar and Podemos 4 seats between them compared with the same votes on one list. Most of those seats go to the PP and the PSOE.

The assumption that the two keep the same ratio in every province is a simplification. If Podemos is relatively stronger in some regions and Sumar in others, the split will fall differently in different provinces. No poll gives provincial figures often enough to measure it.

New parties: SALF

Se Acabó La Fiesta (SALF) did not stand in 2023, so it has no provincial pattern to scale. It is given its national share in every province before the rescale of equation (3). That is the simplest assumption and almost certainly wrong in detail; with SALF polling below 3% nationally it rarely matters, because an even spread at that level wins no seats anywhere.

Regional parties and local lists

ERC, Junts, EH Bildu, the PNV, the BNG, Coalición Canaria and UPN stand only in their own provinces, and proportional swing keeps them there: their national ratio is applied to their provincial shares, and in every other province their share stays at zero. That is right for them, and it is why their seat totals are among the most stable in the forecast.

Local lists that are not in the model sit in "others", which never wins seats. In 2023 no such list won a seat, and the check above is exact. In November 2019 one did: Teruel Existe won a seat in Teruel with 26.7% of the vote. If a local list won seats at the next election, the projection would miss them. In 2023 Soria ¡Ya! took 19.1% of the vote in Soria and Teruel Existe 14.9% in Teruel, both without a seat; a modest swing towards either could change that.

The last seat

For each province the site reports which party takes the last seat, which party is next in line and how much more of the vote it would need. With \(M\) seats, list the quotients of all eligible parties, largest first. The last seat goes to the \(M\)-th quotient, \(q_M\). The party next in line is the one whose next quotient, \(z_{k,p} / (s_{k,p} + 1)\), is largest among those that did not win; it would take the seat if its share rose to

\[ z_{k,p}^{*} = q_M \,(s_{k,p} + 1), \tag{4} \]

so the shortfall shown is \(z_{k,p}^{*} - z_{k,p}\) points, holding the other parties' shares where they are. The provinces where that shortfall is under a point are these:

ProvinceSeatsLast seatNext in linePoints short
Almería6PPPSOE0.15
Bizkaia8PSOEEH Bildu0.20
Barcelona32PPSumar0.24
Palencia3VoxPP0.42
Ceuta1PPVox0.42
Madrid37VoxPP0.46
Pontevedra7PPPSOE0.47
Alicante/Alacant12PPSumar0.53
Zamora3PPVox0.56
La Rioja4PSOEVox0.56
Araba/Álava4PNVPSOE0.58
Navarra5EH BilduVox0.67
A Coruña8PSOEPP0.68
Girona6ERCPP0.69
Asturias7PSOEVox0.79
Lleida4PSOEERC0.94
Live The last seats: provinces where the party next in line is less than a point short Computed from the database, polls to 19 November 2026.

These are the seats the forecast is least sure of, and the simulation (Simulating Spanish polling error: where the chances come from) gives each of them a chance for each party.

The same engine everywhere

One function, SpainForecast.Run, takes a set of national shares and returns the full Congreso: projected shares in every province, the D'Hondt count and the party totals. The projection, the published polls, every past-miss replay (If we're wrong: replaying past Spanish polling misses on today's polls), the reader's own error on the what-if page, the reader's own shares on the prediction page, each of the 2,000 simulated elections and each run of the polling-day forecast go through it. Differences between any two of them are differences in the shares they start from, never in how votes become seats.

What we know about the projection

We do not know as well as we would like. The British seat model was chosen by a backtest (From votes to Commons seats: UK swing models and their test). The same test is possible for Spain, since we hold provincial results for every election since 2004: start from one election, apply the true national shares of the next, and count how many provincial seats come out right under proportional swing, uniform swing or a mixture. It has not been run yet. Until it is, proportional swing is a reasonable default, not a tested choice.

Some weaknesses are clear without a test.

Regions move differently. Spanish elections often have regional swings of their own, in Catalonia and the Basque Country above all. Proportional swing assumes every province moves in proportion to the national change.

Candidates and pacts. Closed lists make individual candidates matter less than in single-seat systems. Joint lists matter more: a pact between two parties in one region, or a regional list standing for the first time, changes that region's count in ways no national swing can show.

Turnout. Each province keeps its 2023 turnout implicitly, because shares are projected, not votes. A change in the balance of turnout between provinces does not change any province's seats, since those depend only on shares within the province.

Apportionment. Seats per province are those of 2023. The next decree will re-apportion them as populations change, moving seats between provinces; the effect on party totals is small but not zero.

Limitations

The legal half is as good as the check: exact for 2023 in the site's code and for every election since 2004 in the pipeline's. The projection half is the weak point. It is untested against a past election, it has no regional layer, it treats the Sumar–Podemos split as the same in every province, it spreads SALF evenly, and it cannot give a seat to a local list. What we don't know yet: limitations of the Spanish forecast collects these points with the rest of the Spanish forecast's weaknesses.

References

Balinski, M. L. and Young, H. P. (1982). Fair Representation: Meeting the Ideal of One Man, One Vote. New Haven: Yale University Press.

Ley Orgánica 5/1985, de 19 de junio, del Régimen Electoral General (LOREG), articles 96, 162 and 163.

Rae, D., Hanby, V. and Loosemore, J. (1971). Thresholds of representation and thresholds of exclusion: an analytic note on electoral systems. Comparative Political Studies, 3(4), 479–488.

Real Decreto 400/2023, de 29 de mayo, de disolución del Congreso de los Diputados y del Senado y de convocatoria de elecciones.

How to cite

Polling Forecast (2026). From votes to seats: D'Hondt in 52 provinces. Retrieved 21 November 2026 from https://www.pollingforecast.com/research/spain-seat-law. Live figures change as new polls arrive; the date above is when the text was last revised.

You may also like