Polling & dynamics

The shape of the race

California’s open primary sends the top two finishers to November regardless of party, and that rule defines the 2026 governor’s contest. As of late May, the field is bunched: no candidate has cleared roughly 22 percent in any recent survey, and the leading candidates sit within a few points of one another. The most recent non-partisan reads place Xavier Becerra (D) narrowly in front in the high teens to about 21 percent, with Steve Hilton (R) and Tom Steyer (D) clustered just behind in the high teens. Katie Porter (D) and San Jose mayor Matt Mahan (D) trail in the high single digits to low teens, and the remaining contenders sit in the low single digits.

Underneath the topline is a structural asymmetry. The Republican vote is concentrated across two candidates, Hilton and Chad Bianco, for a combined share in the high 20s to low 30s. The viable Democratic vote is larger in total — roughly 50 to 60 percent — but split across half a dozen candidates. That fragmentation, more than any single candidate’s strength, is what keeps the second top-two slot unsettled. Aggregators including Race to the WH and RealClearPolling describe the same picture: a tight top tier and an undecided bloc of roughly 10 to 15 percent heading into the final week.

How it moved

The cycle has unfolded in three phases, each tied to a discrete event.

In the first phase, from February into early April, no candidate led. The PPIC survey fielded February 3 to 11 found what it called a five-way tie, with Hilton, Porter, Bianco, Eric Swalwell, and Steyer all bunched between 10 and 14 percent. A March Berkeley IGS poll showed a similar cluster atop a field with a very large share of voters expressing no preference. The two Republicans sat at or near the top largely because they shared the only consolidated lane, while Democrats divided their support several ways.

The second phase began with Swalwell’s exit. He suspended his campaign on April 13 following reporting on misconduct allegations, having polled around 11 percent. An Emerson College poll fielded April 11 to 14, straddling the suspension, caught the field mid-realignment. By the May 9 to 10 Emerson poll, Becerra had risen roughly nine points to 19 percent, which the pollster attributed to his becoming the top choice among Democratic voters. Reporting at the time, including from NBC News, described Swalwell’s departure as opening a lane that Becerra rather than the other Democrats largely filled, aided by labor endorsements and donor relationships that shifted his way. Betty Yee (D) suspended on April 20 while polling in the low single digits; her exit had a smaller and more diffuse effect.

In the third phase, through May, the top tier hardened into a stable order: Becerra in front, with Hilton and Steyer competing for second and the rest in single digits to low teens. Both Swalwell and Yee remain on the printed ballot because ballots were finalized before they withdrew, so a small share of votes cast for them will still be counted.

Reading the pollsters

Not all of the numbers carry equal weight, and the distinction is about who paid for the survey. The load-bearing data come from non-partisan, independently funded pollsters: PPIC, which uses live bilingual phone interviews and tends to show a flatter field with more undecideds; Emerson College, which uses mixed-mode sampling and publishes crosstabs; USC Dornsife, which runs a probability-based panel on a rolling field window; and Berkeley IGS, historically the most accurate close-out poll in California.

Two of the late-cycle surveys were sponsored, and their toplines should be read with that in mind rather than discarded. David Binder Research conducted polls funded by interested coalitions — a labor union in April, a health coalition in May — and each released number favored the sponsor’s preferred candidate, with internals disclosed selectively. The California Democratic Party’s in-house poll, conducted by Evitarus, did not disclose its sample size; a party in-house poll may be framed to motivate turnout, so its exact margins warrant caution. These sponsored surveys are useful for corroborating direction — that Becerra leads and that the second slot is contested — but their precise figures are less reliable than the independent reads. Where the independent and sponsored polls agree on the shape of the race, confidence is higher; where they diverge on a candidate’s exact share, the non-partisan number is the safer anchor.

Caveats

Several factors limit how much any single number should be trusted. The undecided share, 10 to 15 percent in late polls and far higher earlier, is large enough that any of the top five could move several points; in low-information June primaries, late deciders often break toward the perceived front-runner. Vote-by-mail began May 4, so a meaningful share of ballots was cast before the latest polls and arguably before some voters had absorbed the April suspensions, meaning some movement is already locked into returned ballots. Top-two primaries are also intrinsically hard to model because voters can cross party lines, which standard partisan-turnout screens do not capture well. Finally, the single most valuable remaining data point — the final Berkeley IGS poll, expected in the closing days — had not been released as of May 25, and the first county-level mail-ballot return splits will be the earliest real-world signal.