Nearly 150,000 California voters had their mail-in ballots rejected in the June primary, and the number that matters most is not the total. It is 93,479 ballots rejected for lateness, driven by whether the envelope was postmarked in time. State data shows 148,241 mail ballots were rejected, a 1.73% rejection rate of all mail ballots, up from the 2024 primary, when 108,982 mail ballots were rejected, or 1.56%. The jump is also a recognizable escalation from 2022, when 105,818 mail ballots were voided during the primary.

This is a big deal because California already carries a reputation for ponderously slow vote counts, with results sometimes dragging on for weeks or longer. That slowness is often blamed on layers of election changes over the years aimed at increasing turnout but not necessarily improving speed. One of the core mechanics is simple: every voter receives a mail-in ballot that can arrive at an election office seven days late and still be counted, as long as the envelope is postmarked by Election Day. So if you are thinking “the ballots are arriving late,” California’s own breakdown argues otherwise. Voting experts point to postmarks applied after Election Day, making ballots ineligible to be counted even if they show up within that seven-day cushion.

Zoom in on the “lateness” category and the story sharpens. According to the data, late-arriving ballots drove the rejections in June, and lateness accounted for 93,479 rejected ballots, by far the largest category for disqualifying a ballot. That figure rose from about 70,000 in the 2022 primary and about 76,000 in the 2024 primary, though overall turnout in those elections was smaller compared to last June. In other words, the system is not rejecting randomly at higher rates. It is rejecting more of the same problem: postmarks that do not clear the legal deadline.

Kim Alexander, president of the nonpartisan California Voter Foundation, called this out directly, writing in an email that “Ballots rejected due to lateness are caused primarily to being postmarked too late to count, not because they arrived too late to count.” That nuance matters operationally. It changes what officials and the public should try to fix. If ballots are physically arriving within the allowed window but being postmarked late, then the bottleneck is likely upstream in postal processing, not downstream in ballot handling.

Mindy Romero, director of the Center for Inclusive Democracy, a nonpartisan research group, framed the issue as a “demand the question why this is happening” moment, tying the rejection spike to postal procedures and a late surge in ballots during the June race. Many voters, she noted through the surrounding context of the report, waited until the final days to make a decision in a wide-open and chaotic race for governor. That late surge, combined with postal-processing timing, can increase the number of ballots that cross the line between “eligible” and “disqualified” without anyone intuitively realizing where the line is drawn.

The policy and regulatory backdrop is where the pressure is building. In January, a group of mostly Democratic U.S. senators sent a letter to the U.S. Postal Service, alarmed that mail-processing changes might affect postmark dates for mail-in ballots in an election year that will determine control of Congress. The updated agency policy, as described in the source, says postmarks might not indicate the first day the Postal Service received the mail, but rather the day it was handled in one of its processing centers. Those processing centers are increasingly likely to be farther away from certain communities because of recent consolidations within the Postal Service, which could delay postmarks. In other words, the “clock” used by elections may no longer match voters’ expectations about when their ballot was “in the system.”

The USPS dispute is not hypothetical. On Thursday, two Wisconsin Democrats, U.S. Sen. Tammy Baldwin and Gov. Tony Evers, called on the Postal Service to address delays that caused hundreds of ballots to arrive after Election Day in Wisconsin in April. Meanwhile, California election officials earlier this year urged voters to send their mail ballots early, warning that Postal Service changes could lead to slower mail service. And in June, a federal judge halted President Donald Trump’s executive order that sought to create a federal voter list and limit who can receive a mail ballot. Together, these developments show a broader governance fight over how mail voting is administered and verified, not just how votes are counted.

If you are looking for proof that the problem is uneven by geography, the source provides it. In California, the highest mail ballot rejection rate in June, at 3.52%, was in Tulare County, in the agriculture-heavy Central Valley. It was followed by Alpine County, south of Lake Tahoe, and Merced County, also in the Central Valley, which both notched a 3.36% rejection rate. Beyond lateness, other reasons for rejection were also documented: nearly 44,000 ballots were set aside because the voter’s signature did not match the one on record with county officials; over 8,300 ballots were rejected because there was no signature on the returned ballot envelope; and in 743 cases, the voter had already cast a ballot, so the mail ballot was voided. Other reasons included a ballot missing from the return envelope or multiple ballots sent back in a single envelope.

For executives, boards, and investors tracking operational risk in high-stakes systems, this is the second-order lesson: the most consequential failure mode may be timing metadata, not the physical delivery of goods. California may be a national laggard in vote counting, but the June primary shows that even when ballots arrive within counting windows, the “postmark record” can still strand them. After the June primary, President Donald Trump seized on California’s slow-count reputation to renew his criticism of the state’s elections, while the Republican’s Department of Justice launched an investigation into Los Angeles County’s elections. There is no indication of widespread problems with mail voting, and a report by the Brookings Institution published in 2025 found mail voting fraud was minuscule, about four cases per 10 million mail ballots. Still, for peers responsible for compliance-heavy operations, this story is a reminder that when process rules rely on third-party timestamps, disruptions can scale into large, visible outcomes fast.