Who will win Colombia's 2026 presidential election (first round 31 May, runoff 21 June)?

TL;DR

Abelardo de la Espriella is the clear favorite to win Colombia’s June 21 runoff.

Context

The first round is over, so this is no longer a five-way race. The official second-round ballot has only Iván Cepeda–Aída Quilcué and Abelardo de la Espriella–José Manuel Restrepo, and the vote is scheduled for June 21, 2026 (Registraduría, June 5) (registraduria.gov.co).

The CNE’s final first-round scrutiny put De la Espriella first with 10,366,143 votes, or 43.78%, and Cepeda second with 9,703,921 votes, or 40.98%; Paloma Valencia had 1,640,989 votes and Sergio Fajardo had 1,009,346 votes (Infobae, June 4) (infobae.com). Valencia and Fajardo now have only legal-substitution paths, not normal electoral paths.

Evidence

The historical base rate favors the first-round leader, but not enough to settle the question. Since the 1991 Constitution, Colombia has had eight completed presidential elections before 2026. Two were won outright in the first round, and in the six runoffs the first-round leader won four times. The full post-1991 record is:

De la Espriella’s lead is stronger than it first looks. He starts the runoff 662,222 votes ahead, and the largest eliminated bloc is Valencia’s 1.64 million votes. Valencia immediately endorsed De la Espriella (EFE, June 1) (efe.com). Fajardo, with just over 1.0 million votes, issued conditions rather than a clean endorsement for either finalist (El País, June 3) (elpais.com). That makes the obvious transfer arithmetic favorable to De la Espriella: Cepeda can win Fajardo voters and still lose if most Valencia voters follow her.

The post-first-round polls point the same way. I normalize each poll to a two-candidate share, because blank, null, and undecided votes do not elect a president.

The poll average is about 53.5% for De la Espriella on a two-candidate basis. A quality/recency-weighted version that leans more on Guarumo and CNC is lower, about 53.3%. My transfer model from the certified first round lands in the same range: a right-heavy Valencia transfer, a Cepeda-leaning Fajardo transfer, and some blank/abstention among minor-candidate voters gives De la Espriella roughly 53% to 54% of the two-candidate vote.

I convert that to a win probability with a wide error band. The final pre-first-round polling average from AS/COA had Cepeda at 38.0%, De la Espriella at 31.3%, and Valencia at 14.2% as of May 26, while the certified result was De la Espriella 43.78%, Cepeda 40.98%, and Valencia about 6.9% (AS/COA tracker; Infobae, June 4) (infobae.com). That miss was partly late right-wing consolidation, not a simple two-candidate polling error, but it argues against treating a 3–4 point two-candidate lead as safe.

My main calculation is:

$$P(D\ wins)=\Phi\left(\frac{s_D-0.50}{\sigma}\right)$$

where $s_D$ is De la Espriella’s estimated two-candidate vote share and $\sigma$ is the total final-week error in that share. With $s_D=0.534$ and $\sigma=0.039$, the normal model gives about 81%. I keep that near the final number because the polls, first-round vote math, and endorsements all agree.

Legal and administrative risk is small. The first-round scrutiny was closed with De la Espriella and Cepeda certified for the runoff (Caracol, June 4) (caracol.com.co). The Council of State also rejected a demand seeking to knock De la Espriella off the ballot over nationality issues (La Silla Vacía, June 12) (lasillavacia.com). Article 190 creates a replacement path in cases like death, permanent physical incapacity, or another vacancy, but that is a tail risk rather than a live campaign path (Constitute Project, Article 190) (constituteproject.org).

What's non-obvious

The key update is that the right solved its coordination problem before the first round ended. In May, Valencia looked like a possible right-of-center finalist. In the actual count, she fell below 7%, De la Espriella led, and then Valencia endorsed him. Cepeda is no longer facing a divided anti-Petro field.

The other hidden point is that De la Espriella is favored but not safe. The latest CNC poll has him only 52.1% on a two-candidate basis, and the Colombian polling environment just produced a large first-round miss. Cepeda’s plausible path is not persuasion alone. It is a turnout and blank-vote path: convert Fajardo/center voters, keep anti-De-la-Espriella voters from voting blank, and mobilize low-propensity Petro-aligned voters in the final week.

Limitations

The main uncertainty is polling calibration. AtlasIntel has a large sample and caught some of De la Espriella’s rise, but it uses a digital RDR method; Guarumo and CNC are more traditional, but they also missed the first-round shape. I therefore use a much wider error band than sampling error.

I could verify the official ballot and the CNE-reported first-round result through the Registraduría/CNE-linked reporting, but I did not find a clean direct public CNE E-26 table that was easier to cite than the contemporaneous Colombian reports. I also found references to one late outlier poll favorable to Cepeda, but not enough public technical detail to give it meaningful weight.

The small nonstandard tail comes from death, withdrawal, court disruption, replacement under Article 190, a delayed official declaration, or a no-result-by-July-31 edge case. None is likely, but all matter because Paloma Valencia, Sergio Fajardo, and Other are otherwise off the normal ballot path.