Before a single vote is cast in November, each party has to decide who carries its label into the general election. How that happens, who gets to weigh in, and whether the parties even control the choice all vary by state. The structure of a state's primary system shapes who runs, who wins, and which voters get a say in nominee selection. The type of primary is one of the most consequential structural variables in state politics: it determines whether a 35-year incumbent can be picked off by a primary challenger, whether independents can punish a polarizing nominee, and how much room there is for moderates inside each party.
What a Primary Does
A primary election picks each party's general-election nominee. That sounds straightforward, but it papers over a deep ambiguity: is the primary a private party affair (parties choosing their own representatives) or a public election (the state selecting candidates from a pool open to all voters)? Different states have answered that question differently, and the U.S. Supreme Court has weighed in repeatedly, producing the patchwork that exists today.
State legislatures, party committees, and ballot initiatives have all reshaped primary rules over the past three decades. A handful of states still hold conventions or caucuses for some offices. A few have replaced traditional party primaries entirely with all-candidate ballots (Louisiana since 1975, Washington since 2008, California since 2012, Alaska since 2022). The dominant model is still some flavor of partisan primary, but the variations matter.
Primaries vs. runoffs. A primary picks nominees. A runoff is a second round held when no primary candidate clears a threshold (usually 50%+1) in the first round. Nine states (concentrated in the South) require a runoff in at least some primary contexts: Alabama, Arkansas, Georgia, Louisiana, Mississippi, North Carolina, Oklahoma, South Carolina, and Texas. See Runoff Elections for the mechanics by state.
The Six Variants of State Primaries
The National Conference of State Legislatures groups state primary systems into six categories based on who is allowed to vote in each party's primary. The categories are surprisingly granular, and the distinctions matter for both campaign strategy and constitutional analysis.
| Type | Who can vote in a party’s primary? | States |
|---|---|---|
| Closed | Only voters registered with that party. | 8 |
| Partially closed | Each party decides each cycle whether to let unaffiliated or other-party voters in. | 9 |
| Open to unaffiliated only | Unaffiliated voters may choose a party ballot; registered partisans cannot cross over. | 8 |
| Partially open | Any voter, but the voter must publicly declare which party’s ballot they want. | 5 |
| Open | Any voter, no public declaration needed. | 15 |
| Top-two / Top-four | All candidates on one ballot regardless of party; top finishers advance. | 3 (CA, WA, AK) |
Louisiana's nonpartisan blanket primary is structurally similar to top-two but with a wrinkle (a majority winner can be elected outright at the first round). Nebraska's unicameral state legislature uses a nonpartisan top-two model for legislative races only, with party-affiliated primaries for other offices.
Closed and Partially Closed
In a closed primary, only voters who are registered with a party may vote in that party's primary. Independents are excluded entirely (or must change their registration well in advance). Closed primaries reflect the view that parties have a right to control their own nominee-selection process and shouldn't have outsiders deciding their candidates.
Two caveats on this list. Tennessee has no party registration: voters declare a party at the polls when requesting a ballot, which makes the "closed" label technically correct per NCSL classification but functionally permissive, since there's no registration gate to enforce. Wyoming moved to a hard-closed system recently: HB 103 (2023) created a 96-day blackout period before the primary that bans party-switching, and the law has been challenged in state court, where the district court upheld it pending appeal to the Wyoming Supreme Court.
In a partially closed system, the state allows each party to decide each cycle whether to admit unaffiliated voters (and sometimes other-party voters). This gives parties the option to open up when they want broader reach or close down when they want core-voter discipline.
The 1986 U.S. Supreme Court case Tashjian v. Republican Party of Connecticut established that even when state law mandates a closed primary, individual parties have a First Amendment right to open their primaries to independents if they choose. That ruling is what makes partially closed systems constitutionally workable: the party, not the state, holds the final say.
Open and Partially Open
An open primary lets any voter participate in either party's primary without a public declaration of party affiliation. The voter walks into the polling place, picks one party's ballot, and votes. They cannot vote in both. Open primaries reflect the view that primaries are public elections funded by all taxpayers and should be accessible to everyone.
Partially open systems are nearly identical, except that the voter's choice of party ballot is recorded publicly (and in some states, treated as a quasi-registration with that party). The public declaration distinguishes these from fully open systems.
"Crossover voting" in open primaries is real but smaller than its reputation suggests. Academic studies (Sides, Cohen, and Vavreck on "Operation Chaos" in 2008; later work on more recent cycles) find that crossover voting in open primaries typically runs 3 to 20 percent, but strategic "mischief" voting (deliberately picking the weakest opponent for the other side) is hard to detect in the data and rare in practice. Most crossover is sincere: voters who don't strongly identify with one party voting for candidates they actually prefer.
Open to Unaffiliated Voters Only
A distinct intermediate category lets independents pick a party ballot but blocks registered partisans from crossing over. This protects parties from raid-style cross-voting while still giving independents a voice. The category overlaps geographically with New England plus a few western states.
New Mexico is new to this category. Senate Bill 16, signed by Governor Lujan Grisham in April 2025, ended the state's closed-primary system. The semi-open structure takes effect for the June 2026 primary, the first cycle when New Mexico's roughly 320,000 unaffiliated voters can request a major-party ballot without first changing their registration. Sources using older NCSL classifications will still list New Mexico as closed.
Top-Two and Louisiana's Jungle
Louisiana built the original all-candidate primary system. Since 1975, Louisiana has put every candidate on a single ballot regardless of party for most state offices. The first round is held on the federal general election date in November (not in the spring or summer). If any candidate clears 50% in that round, they win outright with no runoff. If no one does, the top two advance to a December runoff. Louisiana's was the first state-level move away from traditional party primaries, and it came out of post-Reconstruction politics specifically: the system was designed to break the lock the state Democratic Party held on every contested office.
Starting in 2026, Louisiana moves congressional, state board of education, public service commission, and state supreme court races to semi-closed partisan primaries, while keeping the jungle system for the state legislature, governor, and most other statewide offices.
The modern top-two primary in California and Washington is the 21st-century variant of the Louisiana model, with one significant structural difference: top-two always sends the top two finishers to a November runoff, even if one of them cleared 50% in the first round. There is no possibility of being elected outright in the primary. The top-two design also handles candidate party labels differently: candidates self-designate a "party preference" on the ballot, but no party formally nominates anyone.
Nebraska's unicameral state legislature has used a nonpartisan top-two system since the chamber was created in 1937. All legislative candidates appear on one ballot without party labels, and the top two finishers advance to the November general. Other Nebraska offices (governor, secretary of state, attorney general, etc.) still use partisan primaries.
Alaska's Top-Four with Ranked Choice
Alaska adopted the most ambitious primary reform of any state in 2020, when voters narrowly approved Ballot Measure 2 (50.55% approval). The new system pairs two structural changes:
- A top-four primary: all candidates on one ballot, top four finishers advance regardless of party.
- A ranked-choice general election: voters rank the four finalists in order of preference. If no candidate has a majority of first-choice votes, the lowest-ranked is eliminated and their voters' second choices are redistributed. The process repeats until one candidate has a majority.
The system was first used in the 2022 special election for Alaska's U.S. House seat (won by Democrat Mary Peltola in a state President Trump carried by 10 points in 2020), and again in the regular 2022 cycle for House, Senate, and state offices. Voters reaffirmed the system in 2024 by defeating a repeal initiative in a near-tie (50.1% No to 49.9% Yes).
Alaska is the only state currently using top-four with ranked-choice tabulation in the general, though Maine uses ranked-choice voting in some federal and presidential elections without the top-four primary structure.
Convention Hybrids
Conventions and party assemblies preserve more party-elite control over nominations than open primaries do. They show up in three distinct roles, and the difference matters for how a state's results are read: in some states a convention is the nomination, in others it merely decides who reaches the primary ballot, and in a few it serves only as a fallback when a primary is inconclusive.
When the Convention Is the Nomination
A handful of states nominate certain statewide offices entirely by party convention, with no primary for those offices at all. South Dakota chooses its governor by primary (with the top two finishers advancing to a runoff if no one clears 35%), but nominates lieutenant governor, attorney general, secretary of state, auditor, treasurer, commissioner of school and public lands, and public utilities commissioner at party conventions held each June, making it one of only a few states that still fills its down-ballot constitutional offices this way. Michigan likewise uses primaries for governor and the legislature but nominates attorney general, secretary of state, the Michigan Supreme Court, and the state education and university governing boards by convention; its parties hold a spring endorsement convention and a later formal nominating convention.
Virginia is the most flexible of all: Virginia law lets each party's "duly constituted authorities" choose the method (primary, convention, mass meeting, or canvass), office by office and cycle by cycle, with narrow protections for incumbents. That is why Virginia parties run conventions for some nominations and primaries for others in the same year, and why the state's frequent legislative special elections are usually settled by party-run canvasses or "firehouse primaries" (single-day, party-run elections held at local polling sites) rather than state-administered primaries.
When the Convention Gates the Primary Ballot
More common is the arrangement where a convention or assembly decides who gets onto the primary ballot, but voters still pick the nominee. An uncontested favorite who clears the convention threshold can become the nominee without any primary appearing at all:
- Utah: the dual-path SB54 system (2014) lets a candidate reach the ballot by party convention or by gathering signatures, or both at once. A candidate who wins 60% of the convention delegate vote is nominated outright with no primary; otherwise the contest goes to a June primary. The U.S. Supreme Court declined to review the law in 2019.
- Colorado: county and state assemblies designate candidates for the June primary (a candidate needs 30% of the assembly delegate vote), with a petition path as the alternative.
- New York: the state party committee's nominating convention designates a candidate, and a contender otherwise reaches the primary ballot by winning at least 25% of the convention vote (or by collecting 15,000 petition signatures statewide).
- New Mexico: a pre-primary convention certifies candidates to the June ballot; 20% of the convention vote secures a spot on the primary ballot, with signatures as the alternative path.
- Connecticut: the party convention endorses a candidate, and a challenger reaches the August primary only by winning at least 15% of convention delegates. If no one clears 15%, the endorsed candidate is the nominee with no primary. This is the same associational logic at issue in Tashjian (below).
When the Convention Is a Fallback
Iowa runs a standard primary but layers on a threshold: if more than two candidates run and none reaches 35% of the vote, the nomination is decided not by a runoff but by a convention of party delegates (or a party committee, depending on the office). The rule is rarely triggered, but it means an Iowa primary plurality does not always settle the nomination. South Dakota applies the same 35% threshold to its governor primary but resolves it with a runoff rather than a convention.
The Constitutional Frame
Three U.S. Supreme Court cases set the boundaries of what states can require parties to do, and what parties can demand for themselves.
California Democratic Party v. Jones (2000)
California voters in 1996 approved Proposition 198, creating a "blanket primary" in which every voter received one ballot listing all candidates from all parties, and the top vote-getter from each party advanced to the general. The state's political parties sued, arguing they were being forced to let non-members pick their nominees.
The U.S. Supreme Court agreed, 7-2 (opinion by Justice Scalia). The blanket primary violated the parties' First Amendment freedom of association by compelling them to accept non-members in their candidate-selection process. The ruling established that primaries are, at their core, party-controlled associational activity, and the state cannot require otherwise.
Washington State Grange v. Washington State Republican Party (2008)
After Jones invalidated California's blanket primary, Washington adopted a structurally different system (I-872, approved by voters in 2004): a top-two primary where candidates self-designate a "party preference" on the ballot. The state parties challenged the system on the same associational grounds that won in Jones.
The U.S. Supreme Court upheld the system, 7-2 (opinion by Justice Thomas). The key distinction: under top-two, the state doesn't nominate party candidates at all. The state just runs a general all-candidate election whose top two finishers move forward. Candidates self-label their party preference, but no party is "forced" to accept any candidate as its nominee. The Court left open the possibility of as-applied challenges if ballots misled voters about party endorsement.
Tashjian v. Republican Party of Connecticut (1986)
Connecticut law required closed primaries. The state Republican Party wanted to open its primary to independents to expand its electorate. The state refused, citing its statute. The party sued on First Amendment grounds.
The U.S. Supreme Court ruled for the party, 5-4 (opinion by Justice Marshall). Just as the state cannot force parties to open their primaries to non-members (the Jones rule, eventually), it cannot prevent parties from doing so. The associational right cuts both ways. This is why "partially closed" systems exist: states have to defer to each party's choice each cycle.
The 2024 Reform Wave (and Voter Backlash)
2024 was the most active year for state primary reform in decades, and the most lopsided defeat. Five states had primary-restructuring measures on the November 2024 ballot, and voters rejected every single one, often by punishing margins.
| Measure | Proposal | Result |
|---|---|---|
| AZ Prop 140 | Single all-candidate primary with possible RCV general | Defeated · 1,284,176 Yes / 1,823,445 No (41.0%) |
| CO Prop 131 | Top-four primary with RCV general | Defeated · 1,385,060 Yes / 1,595,256 No (46.5%) |
| ID Prop 1 | Top-four primary with RCV general | Defeated · 269,960 Yes / 618,753 No (30.4%) |
| NV Question 3 | Top-five primary with RCV general (second of two required passages) | Defeated · 664,011 Yes / 747,719 No (47.0%) |
| SD Amendment H | Top-two open primary (no RCV) | Defeated · 141,570 Yes / 270,048 No (34.4%) |
Several patterns held across the defeats. Opposition campaigns successfully framed the proposals as confusing or as elite-driven projects rather than grassroots reforms. Established party committees and most major-party officials lined up against the changes. Ranked-choice voting in particular drew opposition from voters who saw it as adding complexity rather than simplifying ballots.
Nevada's Question 3 is a special case: it had passed in 2022 (53-47%), but Nevada constitutional amendments require two consecutive passages by voters, so the 2024 result killed it. Idaho's nearly 40-point margin was the largest of the defeats and effectively ended ranked-choice voting as a near-term reform target in red states.
The 2024 results reframe the primary-reform debate. Voters in five geographically and politically diverse states rejected what reformers were selling, even where the case was made carefully and the money was substantial. Whether the rejection was about the specific proposals (RCV's complexity, top-four's novelty) or about the broader theory that primary reform improves outcomes is the open question for the next cycle. Alaska's reaffirmation of top-four-plus-RCV in 2024, by defeating Ballot Measure 2 (the repeal initiative) by a razor-thin 50.1-49.9% margin, is the one reform that survived the wave.
Does the Type Matter?
The most common political argument for primary reform is that open systems and top-two systems produce more moderate, less polarizing candidates. The empirical record is mixed.
On polarization broadly. A widely cited 2014 paper by McGhee, Masket, Shor, Rogers, and McCarty in the American Journal of Political Science, "A Primary Cause of Partisanship? Nomination Systems and Legislator Ideology," found that primary openness has "little, if any, effect" on the ideological extremism of nominated candidates. Later work has been more nuanced. Christian Grose's 2020 study in the Journal of Political Institutions and Political Economy, "Reducing Legislative Polarization: Top-Two and Open Primaries Are Associated with More Moderate Legislators," finds that top-two primaries in California and Washington produce measurably more moderate legislators than closed-primary states. The University of Chicago's Effective Government Initiative reviews the broader literature and concludes the moderation effect of top-two is real but small. Advocacy claims that "open primaries reduce polarization" overstate the broader-open-vs-closed evidence; the specifically-top-two evidence is stronger.
On independent-voter access. The exclusion question has a sharper answer. According to Gallup's January 2025 polling, 43% of Americans identify as political independents, with 28% Democrat and 28% Republican. In closed-primary states, the largest single bloc of voters is excluded from the primary entirely. (Worth a caveat: most self-identified independents lean reliably toward one party in their actual voting behavior, so "43%" overstates the swing-voter share. But the exclusion-from-primaries point holds regardless.) That fact drives much of the independent-voter advocacy for primary reform, regardless of whether it changes who gets nominated.
On crossover voting. The empirical record on whether voters strategically cross over to "raid" the other party's primary is consistent: it happens, but it's rare and mostly sincere. The 2008 "Operation Chaos" episode (Rush Limbaugh's call for Republicans to vote in Democratic primaries to extend the Clinton-Obama contest) was widely covered at the time, but careful analysis of Indiana, North Carolina, and Pennsylvania exit polls and registration data found no measurable Limbaugh-driven crossover. When voters cross over, they typically do so because they actually prefer a candidate in the other party, not because they're trying to game the outcome.
On candidate recruitment. Closed primaries reward candidates who can mobilize base party voters; open and top-two primaries reward candidates who can reach a broader electorate. For incumbents in safe partisan districts, an open or top-two primary is genuinely a tougher renomination environment than a closed one, because moderates and independents can vote against them. This is the structural intuition behind both the most enthusiastic advocacy for primary reform and the strongest resistance from party committees.
On what voters actually do at the ballot box. The 2024 results discussed above are the freshest evidence on whether voters themselves want reform. Across five states with diverse politics, the answer was a clear no. That doesn't settle whether reform should happen, but it sharply qualifies the claim that there's pent-up demand for it.
Where to go next. See Runoff Elections for what happens when a primary doesn't produce a majority winner, Legislative Districts for how the geographic structure interacts with primary mechanics, or any state's profile page for the specific rules in play.