Maine’s ranked-choice voting (2016 to present)

Maine’s ranked-choice voting (RCV) was adopted by citizen referendum in November 2016 and now applies to state and federal primaries, congressional general elections and presidential elections, but not to general elections for governor or the Legislature, which the Maine Supreme Judicial Court advised in 2017, and again in 2026, would conflict with the state constitution’s plurality provisions. In the 2018 Second Congressional District race, Bruce Poliquin led the first round 134,184 to 132,013, but after tabulation Jared Golden won 142,440 (50.62%) to 138,931 (49.38%) of ballots still active, with 14,706 ballots (4.97%) exhausted. A federal judge rejected the challenge to that count.

Ranked-choice voting lets voters rank candidates: first choice, second choice, and so on. If nobody gets more than half of the first choices, the last-place candidates are removed, and their voters’ next choices are counted. Maine voters chose this system in 2016. Lawmakers tried to delay it, voters overruled them, and the state’s highest court said it cannot be used for some state offices. In one close congressional race, the person who was ahead on first choices lost once the rankings were counted, which led to a lawsuit that failed.

For Alaska’s related system, see Alaska’s top-four primary. For British Columbia’s attempt to adopt a ranked system, see British Columbia’s Citizens’ Assembly. For ranked ballots as a research topic, see collective priorities.

Defined terms

Maine’s ranked-choice voting uses several technical terms.

• Ranked-choice voting (instant-runoff): voters rank candidates; if no candidate has a majority of active ballots, the last-place candidates are eliminated and their ballots transfer to the next ranked candidate still in the race.⁠¹
• Batch elimination: removing at once all candidates who cannot mathematically win, as the Secretary of State did for two candidates in 2018.⁠¹
• Overvote: ranking more than one candidate at the same level; undervote: leaving a ranking blank.⁠¹
• Exhausted ballot: a ballot that no longer counts for a remaining candidate, because it was blank, had disqualifying errors, or ranked no remaining candidate. Exhausted ballots are not part of the denominator for the majority.⁠^1,8
• People’s veto: a Maine citizen referendum to reject a law passed by the Legislature.⁠³
• Plurality provisions: sections of the Maine Constitution providing that the governor, state senators and state representatives are elected by a plurality of votes.⁠²

What happened

Maine’s ranked-choice voting has a dated legal history.

• November 2016. Maine voters approved ranked-choice voting by referendum.⁠³
• May 23, 2017. In an Opinion of the Justices requested by the Senate, the Supreme Judicial Court answered that the Ranked-Choice Voting Act conflicts with the Maine Constitution for general elections for state representatives, state senators and governor, because multi-round tabulation would not declare a plurality winner “if there are more than two candidates”.⁠²
• October 2017. The Legislature passed a bill to delay implementation.⁠³
• June 12, 2018. Voters approved a people’s veto that vacated the delay law and applied RCV to state and congressional primaries and congressional general elections; with 77% of precincts reporting, unofficial results showed 127,048 Yes to 106,607 No.⁠³ The same day was Maine’s first statewide RCV election.⁠⁴
• November 6, 2018. Second Congressional District election: no candidate won a majority of first choices.⁠¹
• November 26, 2018. The Secretary of State certified the tabulated result, Golden 50.62%, Poliquin 49.38%.⁠¹
• December 13, 2018. Judge Lance Walker of the U.S. District Court for the District of Maine denied the challenge in Baber v. Dunlap and entered judgment for the defendants.⁠¹
• September 22, 2020. The Supreme Judicial Court unanimously held that a Republican-led people’s veto petition against extending RCV to presidential elections lacked enough valid signatures, after ruling that signature gatherers’ residency requirements did not violate the First Amendment; 988 signatures were invalidated against a requirement of 67,067.⁠⁵ Maine became the first state to use RCV for president.⁠⁵
• April 6, 2026. In a non-binding advisory opinion on LD 1666, the court advised that the constitution does not permit expanding RCV to state general elections.⁠^6,7

How it worked

Maine’s ranked-choice voting is administered centrally by the Secretary of State, which makes it different from systems counted entirely at the local level.

Ballot. In the 2018 Second District race, the ballot listed four candidates plus a write-in line with five ranking columns; the court noted 120 possible full orderings when a write-in was named and every candidate ranked.⁠¹

First round. The Secretary of State gathered the ballots and tabulated first choices: Poliquin 134,184, Golden 132,013, Tiffany Bond 16,552, William Hoar 6,875, total 289,624.⁠¹ In this round 6,453 ballots (2.28%) were exhausted, mostly for undervotes.⁠¹

Later rounds. Because no candidate had a majority, Bond and Hoar were batch-eliminated and their ballots transferred where they ranked Golden or Poliquin; a ballot that left two sequential rankings blank before naming either was exhausted.⁠¹ Another 8,253 ballots were exhausted, for 14,706 in total (4.97% of all ballots cast). Of these, 5,582 were completely blank, 3,957 ranked only Bond, and 1,939 ranked only Hoar.⁠¹

Scope. Under current law, RCV is used for all congressional elections, state primaries, presidential primaries from 2024 and presidential general elections from 2020, but not for general elections for governor and the Legislature.⁠^5,7

Evidence: measured results

Maine’s ranked-choice elections have been measured in official counts and in ballot-level analysis.

2018 Second District:⁠¹

Ballot-level behavior in the June 2018 gubernatorial primary. Drew Kurlowski of the MIT Election Data and Science Lab analyzed 132,250 ballots with 17,513 unique combinations: about 12% ranked one candidate, 10.3% ranked exactly two, and nearly 30% ranked all seven.⁠⁸ He found that 0.7% of ballots had a double undervote that exhausted them, 1.4% contained an overvote (0.4% exhausted by it), 5,535 ballots (4.2%) were completely blank, 10.7% of ballots were ultimately exhausted due to undervotes, and only 265 ballots were exhausted because every ranked candidate had been eliminated.⁠⁸

Participation. FairVote, an organization that advocates RCV, reported that on June 13, 2018, nearly 50,000 more voters had voted on the RCV question than in both parties’ gubernatorial primaries combined.⁠⁴ That compares a referendum with partisan primaries and is not a measure of RCV’s effect on turnout.

Failures and criticisms

Maine’s ranked-choice voting has been criticized and defended, and each position is attributed here.

From opponents: “majority” is a misnomer. The Alaska Policy Forum, in a 2020 report written with the Maine Policy Institute, argues that because exhausted ballots leave the denominator, Golden won with 49.18% of total votes cast, making “majority” “a misnomer”; it reports that across 96 RCV races the winner lacked a majority of all ballots cast in 61% of cases, and that an average of 10.92% of ballots were exhausted by the final round.⁠⁹ It also says Maine proponents had to “publish a 19-page instruction manual” to help voters.⁠⁹

From the challengers in court. Poliquin and other voters argued that RCV was unconstitutional on its face and as applied, violated the Voting Rights Act, used confusing ballots and instructions, and diluted their votes.⁠¹ Judge Walker rejected the claims, writing that such criticism “falls short of constitutional impropriety” and that “A majority of Maine voters have rejected that criticism”.⁠¹ He found that plaintiffs had not shown that a similar volume of votes is not also invalidated in plurality elections.⁠¹

From supporters: the system did what it was designed to do. Supporters argue the 2018 result reflected the preferences of more voters than a plurality win would have, since Bond and Hoar voters who ranked a finalist were counted. The court’s data show that many exhausted ballots were blank or ranked only an eliminated candidate,⁠¹ which supporters read as voter choice rather than system failure.

From the constitution. The Supreme Judicial Court’s 2017 and 2026 opinions mean Maine uses two counting rules depending on office.⁠^2,7 Supporters of RCV, including the League of Women Voters of Maine, “strongly disagree” with the 2026 opinion;⁠⁶ others view the limit as the constitution working as written. The 2026 opinion is advisory and does not affect current uses.⁠⁷

From both sides on process. The Legislature’s 2017 delay, reversed by voters in 2018,⁠³ and the 2020 dispute over petition signatures.⁠⁵ show how the rules for changing voting rules became contested in their own right.

Competing interpretations

Observers read Maine’s experience in at least three ways.

A successful reform. On this reading, voters approved RCV in 2016 and reaffirmed it in 2018, a 2020 repeal petition failed to qualify, courts upheld it where the constitution allows, and error rates are low⁠^3,5,8

A flawed reform imposed by narrow votes. On this reading, exhausted ballots, complexity and the 2018 reversal of the first-round leader show RCV substituting a manufactured majority for a real one.⁠⁹

A rule caught between two sources of authority. On this reading, the central lesson is constitutional: a citizen initiative can change statutes, but a state constitution’s plurality provisions still govern some offices, producing a split system.⁠^2,7

What transferred and what did not

Maine’s adoption influenced other states, and some features did not carry over.

Transferred. Alaska adopted ranked-choice general elections by voter initiative in 2020, combined with a top-four open primary;⁠¹⁰ see Alaska’s top-four primary.

Did not transfer within Maine. RCV did not extend to state general elections.⁠^2,7

Not established. Whether RCV changed campaign tone, candidate entry or turnout in Maine was not established by sources verified for this page.

Application to this framework

Maine’s ranked-choice voting bears on several parts of My Vote Forward.

• Rankings carry more information than single marks. The framework uses ranked priorities; see priorities and collective priorities. Maine shows both the value of rankings and the reporting questions they raise.
• Report every denominator. Golden’s 50.62% of active ballots and 49.18% of all ballots, as the Alaska Policy Forum computes it,⁠^1,9 are both true. The framework asks releases to state which base they use; see measurement.
• Publish round-by-round results. Maine’s central tabulation produced documented rounds that a court could examine.⁠¹ See audits and verification.
• Design for incomplete input. Many voters rank one candidate.⁠⁸ Any ranking interface should treat partial rankings as legitimate, not as errors.
• Rule changes need stable authority. The sequence of initiative, delay, veto and court opinions shows why the framework separates who decides the rules from who participates; see governance.

Unresolved questions

• Does ranked-choice voting change turnout, candidate behavior or campaign tone in Maine? Verified evidence for this page does not settle it.
• Should winners be described by share of active ballots or share of all ballots cast?
• How many voters who ranked only one candidate did so deliberately, and how many misunderstood the ballot?
• Will Maine amend its constitution to allow RCV in state general elections, and would voters approve?
• How should ballot instructions be tested so that error rates fall further?