Alaska’s closely contested 2026 U.S. Senate election has two very different sets of numbers: a narrow Democratic lead in reported first-choice votes and an unofficial ranked-choice simulation showing Republican incumbent Dan S. Sullivan ahead within a smaller pool of ballots. They are not competing election calls. They measure different things — and neither determines the winner.
- How ranked-choice voting works in Alaska
- What the reported first-choice votes show
- Why a preliminary simulation shows a different picture
- Why the two Dan Sullivans matter
- What is missing from the preliminary calculations?
- When will Alaska complete the ranked-choice count?
- How Votendi will report the result
As of 7:26 p.m. Eastern on November 4 (1:26 a.m. CET on November 5), Decision Desk HQ reported 102,962 first-choice votes for Democrat Mary Peltola and 102,325 for Sullivan, a Peltola lead of 637 votes. The race remains uncalled, and thousands of eligible ballots have yet to be incorporated into the final count.
To understand why the first-choice margin is not necessarily the final margin, it helps to look at Alaska’s distinctive election system — and at exactly what analysts have been able to calculate so far.
How ranked-choice voting works in Alaska
Alaska uses a top-four primary followed by a ranked-choice general election. Voters can rank as many or as few of the general-election candidates as they wish, from first choice to their next preferences. They are not required to rank everyone.
The count proceeds in rounds. If a candidate receives more than half the valid first-choice votes, that candidate wins without additional rounds. Otherwise, election officials eliminate the candidate with the fewest votes. Ballots for the eliminated candidate are transferred to each voter’s next-ranked candidate who is still in the contest. The elimination-and-transfer process continues until two candidates remain, and the candidate with more continuing votes wins.
For example, imagine a voter ranks Republican Gerald Heikes first, Sen. Dan S. Sullivan second and Mary Peltola third. That ballot initially counts for Heikes. If Heikes is eliminated while both major contenders remain, the same single vote transfers to the senator. It is not counted twice, and it does not automatically follow party affiliations.
A voter may rank only one candidate. If that candidate is eliminated and no further usable preference remains, the ballot becomes inactive (sometimes called exhausted) for subsequent rounds. Alaska also has rules for overvotes and skipped rankings. The Alaska Division of Elections explains those rules in its official voting guide.
What the reported first-choice votes show
| Candidate | First-choice votes |
|---|---|
| Mary Peltola (D) | 102,962 |
| Dan S. Sullivan (R, incumbent) | 102,325 |
| Peltola’s lead | 637 votes |
Source: Decision Desk HQ, updated November 4, 2026 at 7:26 p.m. ET. These are first-choice figures, not a completed ranked-choice result. Additional candidates also received votes.
These figures tell us which of the two leading candidates had more reported first preferences in that DDHQ snapshot. They do not tell us how ballots for lower-ranked candidates will be distributed, or how still-outstanding ballots will affect the count.
Why a preliminary simulation shows a different picture
On November 4, the Alaska Beacon analyzed an anonymized cast-vote record containing rankings on 204,530 ballots. In that incomplete data set, its simulation calculated that, after the two lower-placed candidates are eliminated, Sullivan would gain 2,728 transferred votes and Peltola would gain 1,868 — a net transfer advantage of 860 for the incumbent.
| Transferred votes in Beacon sample | Votes gained |
|---|---|
| Dan S. Sullivan (R) | 2,728 |
| Mary Peltola (D) | 1,868 |
| Difference in transfers | Sullivan +860 |
Alaska Beacon analysis of an incomplete November 4 cast-vote record, not an official tally or a projection of the final result.
Separately, the electoral-reform organization FairVote ran a preliminary ranked-choice tally on approximately 203,000 recorded ballots. Its simulation showed Sullivan 51% to Peltola 49% in the ballots analyzed. FairVote stressed that this file did not cover every ballot already represented in the wider first-choice totals: the included records even showed Sullivan leading in first choices, while the broader published tally showed Peltola ahead.
These figures cannot simply be combined. Subtracting the Beacon’s 860-vote transfer advantage from DDHQ’s 637-vote lead would mix different snapshots and different sets of ballots. That arithmetic would not produce a valid ranked-choice result.
Why the two Dan Sullivans matter
Four candidates appeared in the Senate contest, including incumbent Dan S. Sullivan and another contender, Daniel J. Sullivan Jr., who appeared without a party affiliation. Republican Gerald Heikes also ran.
Ballots for the second Sullivan do not automatically become votes for the senator, despite the shared surname. They transfer according to the actual rankings recorded by each voter. FairVote found that, within its partial data, a larger share of Heikes voters transferred to the incumbent, while voters who ranked Daniel J. Sullivan Jr. first were more likely to transfer to Peltola than to the senator.
The namesake candidacy has also prompted political and legal disputes over possible voter confusion. Allegations about the recruitment of the second Sullivan have been contested, including by Peltola. Those disputes are separate from the mechanical rules of vote transfer. For background, see Votendi’s earlier reporting on the namesake controversy.
What is missing from the preliminary calculations?
Crucially, the initial digital cast-vote record did not contain every ballot included in the statewide first-choice tally. The Beacon noted that roughly 11,000 ballots already reported in first-choice results were missing from its source file, including votes from 131 precincts without scanners, many in rural Alaska. Additional early, absentee and other eligible ballots were still awaiting counting.
That means a simulation may correctly describe the ballots in its file without describing the complete electorate. New batches could alter both the first-choice standings and the way later rankings affect the final rounds. No campaign or independent analyst has an official final ranked-choice result at this stage.
When will Alaska complete the ranked-choice count?
The Division of Elections plans additional reporting on November 10 and November 13, with its official ranked-choice tabulation scheduled for November 18, 2026, after eligible absentee and other outstanding ballots are processed. Certification is targeted for November 25. These dates are subject to the state’s election administration and any legally required follow-up.
Alaska’s long counting period is not simply caused by ranked-choice voting: the state accepts eligible late-arriving ballots, including from remote communities, under deadlines that predate the current voting system.
How Votendi will report the result
Three labels are essential to keep separate: first-choice leader (reported votes at a specific time), preliminary ranked-choice simulation (an unofficial calculation from a specified ballot sample), and official ranked-choice result (the state tabulation after outstanding ballots are counted).
Until the state completes the count or a reliable election desk calls the race, the Alaska Senate contest should remain uncalled. Follow the Votendi 2026 U.S. midterm live results and the Alaska Division of Elections for updates.
Sources and methodology: Decision Desk HQ (November 4, 7:26 p.m. ET); Alaska Beacon’s November 4 RCTab analysis; FairVote’s November 4 preliminary cast-vote-record analysis; Alaska Division of Elections rules and counting schedule. All provisional vote totals and simulations are time-stamped and should not be treated as a certified result.
