The Senate Agreement: Vox Gives Ground, the PP Gains Fifteen Senators and the Three-Fifths Majority Comes Closer

The PP and Vox have reached a Senate agreement covering 23 provinces. Vox will not field a list in six, the PP will not in three, and Vox will run a single candidate in another fourteen. With today's forecast, the PP goes from 128 to 144 senators and Vox from 9 to 10. The probability that the PP has an absolute majority on its own rises from 39% to 75%, and the probability that the PP and Vox together reach a three-fifths majority of the chamber rises from 14% to 40%. Almost all of the effect comes from the arithmetic of the limited vote, not from voter discipline.

26 October 2026 · Miguel · Leer en español

Update, October 27. In response to the PP–Vox agreement, the Comuns (Frente Amplio) have withdrawn their Senate lists in Girona, Lleida and Tarragona and are asking their voters to back the PSC, which in return will propose a Comuns member as a regionally appointed senator. The model now includes this. With the October 27 forecast the effect is small: the PSC is more likely to secure its three senators in those provinces (in Tarragona, from 61% to 75% of simulations), the left gains a little over half a senator on average, and the probability of a PP–Vox three-fifths majority barely changes (from 47.3% to 46.6%). The figures in the article are from the October 26 forecast.

What they signed

The agreement that the PP and Vox announced on Monday, the day before the candidacy deadline, is not a coalition: each party keeps its own name and the pact is limited to the Senate. It has three parts:

• Vox does not field a list in six provinces: Las Palmas, Santa Cruz de Tenerife, Cádiz, Valencia, Castellón and Baleares. In Baleares, the PP will include on its list an independent proposed by Vox.
• The PP does not field a list in three: Girona, Tarragona and Gipuzkoa. In the two Catalan provinces it has never had a senator, and in Gipuzkoa it has not had one for 26 years.
• Vox fields a single candidate, instead of three, in fourteen: Sevilla, Huelva, Asturias, Badajoz, Álava, Navarra, Cáceres, Huesca, Zaragoza, Alicante, Granada, Cuenca, Pontevedra and Jaén.

It is a response to the joint list that the left agreed earlier for the Senate. According to the model, the joint PSOE–Frente Amplio list will run in almost every province.

The arithmetic of the limited vote

Four senators are elected in each mainland province, and each voter can mark up to three names. The four most-voted candidates win the seats, whatever their party. In practice, the most-voted list takes three senators and the second takes one, even if it gets almost the same number of votes. In 2023, in 34 of the 47 four-senator provinces the split was three for the PP and one for the PSOE, and in seven others it was the PSOE that took three.

That is why it matters so much who comes first in each province. With separate lists, the right-wing vote is split between PP and Vox candidates, and the joint PSOE–Frente Amplio list beats the PP in quite a few provinces. The agreement acts on that split in three ways:

• Where Vox does not run, a good part of its voters will vote for the PP.
• Where Vox runs a single candidate, its voters mark that candidate and have two marks left over. The agreement asks them to give those to PP candidates. Each PP candidate thus adds up the votes of his or her own party and part of Vox’s.
• Where the PP does not run, its vote can go to Vox, which gains chances in Tarragona and Girona.

In theory, Vox’s single candidate could win the fourth seat and leave the left without a senator in that province. To do so, it would need more votes than the top candidate of the joint list, and in almost no province is that likely.

The effect on the chamber

With today’s polls and the agreement, the PP gets a median of 144 senators counting the appointed ones (between 105 and 161 in 95% of simulations), against 128 with separate lists. Vox goes from 9 to 10 and the left from 107 to 91. Comparing each simulation with itself, with and without the agreement, the PP gains 15 senators on average. The joint PSOE–Frente Amplio list loses 12 and the PSOE, where it runs alone, another 2. ERC, Junts, Bildu and Coalición Canaria together lose about one among the four. Vox gains 0.1.

The last row is the most revealing. If the polling average were right, without the agreement the PP and Vox would almost never reach three-fifths (1%). With it, they land at a median of 160 senators, one short of the 161 needed, and reach it in 46% of simulations. The agreement turns a nearly impossible supermajority into one within reach.

Province by province

Where Vox runs a single candidate, the PP goes from about two senators to almost three. Vox’s candidate almost never wins: he or she does so in 11% of simulations in Granada, 9% in Sevilla and 8% in Zaragoza, and in 6% or fewer in the rest. The right takes all four seats just as rarely. In Álava and Navarra the agreement barely changes the result: the PP gains a little, but the left and the nationalists remain ahead.

Where Vox does not run, the PP gains most in Valencia and Castellón, where the joint list was ahead: it goes from 1.4 to 2.6 senators in Valencia and from 1.6 to 2.8 in Castellón. In Cádiz it secures the third seat (from 2.1 to 2.9), which was previously contested with the left and Vox. On the single-senator islands the change is all or nothing. The PP becomes the favorite in Menorca, Formentera, Lanzarote and Fuerteventura and secures Eivissa. In the province of Santa Cruz de Tenerife nothing changes, because Vox has barely any vote there.

Where the PP does not run, Vox gains a little. In Tarragona its average goes from 0.3 to 1.2 senators and it wins at least one in 86% of simulations. In Girona it does so in 45%. In Gipuzkoa the agreement changes nothing.

Vox gives up seats, but gains chances

Seat by seat, Vox gives up a lot. It stops running in Cádiz and Castellón, where it had some chance with separate lists. With a single candidate in fourteen provinces, its chances there fall to half or less. What it gains in return are Tarragona and Girona. In total, Vox’s median goes from 9 to 10 senators, but comparing each simulation with itself it gains barely 0.1 on average: what it adds in Catalonia it loses elsewhere. Nor is a parliamentary group of its own, which requires ten senators, assured: with the agreement it would have one in 57% of simulations (50% without it). According to the PP, the pact includes no commitment on this.

The point of the pact lies not in the number of senators for each party, but in what the bloc can do with three-fifths:

• With an absolute majority (134) the Senate can veto laws, although the Congress can lift the veto, and it approves measures under Article 155 of the Constitution.
• With three-fifths (161) the Senate nominates four of the twelve Constitutional Court justices and elects ten of the twenty members of the General Council of the Judiciary; the Congress elects the other ten.

Today, that threshold can only be reached through an agreement between the two big parties. With the pact and current polls, the PP and Vox would reach it on their own in four out of ten simulations.

What matters most is not discipline

A reasonable question is how many voters will follow instructions. Will the Vox voter mark two PP candidates? Will the PP voter in Tarragona vote for Vox? The model treats this as one more uncertainty:

• Vox voters who vote PP where Vox does not run: between 55% and 85%.
• PP voters who vote Vox where the PP does not run: between 35% and 70%.
• Vox voters who give their two spare marks to the PP where Vox runs one candidate: between 50% and 90%.

The result depends little on those assumptions. Between the simulations with the lowest discipline and those with the highest, the PP and Vox total changes by less than two senators on average, and the probability of three-fifths goes from 39% to 41%. The agreement works mainly because it removes candidacies: fewer right-wing lists competing for the same votes. Whether voters follow them to the letter matters much less.

What does matter is the vote. The difference between the full forecast (40% for three-fifths) and the scenario in which the polls are right (46%) shows that the result still depends on how PP and Vox voters cast their ballots on November 29 and on how much the left’s joint list gets. The PP’s absolute majority on its own, by contrast, is likely in almost any scenario (75%; 98% if the polls are right).

What remains to be known

• The final candidacies. The deadline expires tonight, and the Electoral Board will proclaim them around November 3. The agreement may have fine print we do not know: who the independent in Baleares is, who will go second or third on the PP list, whether there are more provinces or whether any falls through. The list of provinces has itself already changed once: early reports put Lleida, rather than Girona, among those the PP is dropping. Until then, the Senate forecast uses the agreement as announced.
• The left. The model assumes a joint PSOE–Frente Amplio list in almost every province. Any change in that offer would move the result as much as the right’s pact.
• The appointed senators. The 58 senators chosen by the regional parliaments are held fixed (23 from the PP, 19 from the PSOE and 5 from Vox, among others). If any parliament renews its appointment before the chamber is constituted, the split may shift by one or two.
• The Congress. This agreement concerns only the Senate. What Frente Amplio does with its Congress candidacies we will learn from the lists, and we will analyze it then.

How we calculated it

• Senate forecast. The vote shares in each province come from the simulations of the October 26 Congress forecast, corrected for how far the Senate vote deviated from the Congress vote in 2019 and 2023, island by island where needed. Each list fields as many candidates as the law allows. The votes of each candidate within their list are taken from how they were distributed in 2019 and 2023, with a small random error. The most-voted candidates win.
• The agreement. Where Vox does not run, a fraction of its vote, drawn between 55% and 85% in each simulation, goes to the PP. Where the PP does not run, between 35% and 70% of its vote goes to Vox. Where Vox runs one candidate, all of its vote goes to that candidate, and between 50% and 90% of its voters spread their other two marks among the PP’s three candidates. The rest, in each case, abstain or vote for other lists.
• Comparison. The two versions, with and without the agreement, use the same 20,000 vote simulations, so the difference between them is the effect of the agreement and not noise. “If the polls are right” repeats the calculation without the polls’ common error.
• Source of the agreement. El Mundo, October 26, 2026, with statements from the PP and Vox.
• Code. senado_pronostico.py (scenario acuerdo_26_10, now the Senate base scenario) and articulo_senado_acuerdo.py, in the altnalysis repository.