ProPublica finds late-ballot rates doubled or tripled in rural counties with USPS cuts

USPS says routing measures should minimize Nov. 3 delays

In recent weeks, concern about disruptions to mail voting has focused almost entirely on President Donald Trump’s now-shelved proposal to sharply restrict who could vote by mail. ProPublica’s analysis identifies a separate risk with national scope: delays resulting from U.S. Postal Service cost-cutting have endangered the validity of ballots cast in rural areas, not only in Washington, but across the country. In 2026 primary data, late-ballot rejection rates in counties where services were cut doubled or tripled from four years earlier.

The effects were clearest in five states that provide detailed mail-ballot data: Washington, California, Ohio, North Carolina and Oregon. All have or recently had grace periods allowing ballots to count after Election Day if they were postmarked by the deadline. ProPublica found that late-ballot rejection rates doubled or tripled from 2022 in counties where Postal Service services were cut, dwarfing any increases in counties where pickups remained unchanged. In urban Spokane, the number of ballots rejected for being late declined.

In early August, election workers north of Spokane, Washington, went across the street to the post office three days after the primary, expecting no more than a few dozen late mail-in ballots. They found more than 500 envelopes, each containing a vote that would not count. Stevens County Auditor Lori Larsen, a Republican who oversees voting in the county, called the sight “jarring.” Across 11 predominantly rural eastern Washington counties, nearly 2,000 ballots arrived late in 2026, compared with about 600 in the 2022 midterm primary.

In the 30 mostly rural Washington counties most affected by regional transportation optimization, or RTO, the share of ballots postmarked late more than tripled from the 2022 primary, ProPublica found. In Garfield County in southeastern Washington, 19 ballots—nearly 1 in 10 cast—had late postmarks in 2026, compared with one in 2022. Pend Oreille County in northeast Washington along the Idaho panhandle rejected 6% of ballots for arriving late, up from 1%, while Stevens County’s late-postmark rejection rate rose to 8% from 1%.

Pend Oreille’s longtime auditor, Marianne Nichols, said, “It was pretty awful telling people we couldn’t count their votes.” “They were quite upset because they felt like, ‘Why? This was in the mailbox on election day, the post office picked it up and had it.’ They felt like it should have counted.” Steve Hutkins, a retired professor who analyzes the effect of Postal Service changes on Savethepostoffice.com, also found more ballot rejections during school-board and other local elections earlier this year in Washington locations with RTO. The patterns “set off the alarm bells,” he said.

Over the past three years, as mail volume continued to plummet and annual deficits ballooned, the Postal Service began consolidating first-class processing at larger regional plants instead of smaller hubs closer to local post offices. In a switch that mostly took effect after the 2024 election, the agency cut mail-pickup frequency in half at many offices 50 miles or more from the processing plants, reducing pickups from twice daily to once.

ProPublica’s map distinguished planned reductions, reductions already in place, areas with no reduction planned and locations without data. It showed reduced pickups in ZIP codes more than 50 miles from a sorting facility; areas not subject to the plan were mostly in or near large cities. The map named Minneapolis, Boston, Detroit, Chicago, New York, D.C., St. Louis, Greensboro, Atlanta, Jacksonville, Miami, Seattle, San Francisco, Denver, Los Angeles, Albuquerque, Phoenix, San Diego, Dallas, Houston and San Antonio.

The changes often added a day or more to delivery time and pushed back postmarking because mail was not postmarked until it reached a regional center. A ballot deposited after the daily pickup, or shortly before Sunday or a holiday, could take longer. If pickups had been reduced, the postmark could show a later date than the day the ballot was deposited, which can matter for a ballot subject to a deadline.

Across the five states, the delays had their most substantial effect in areas that voted for Donald Trump in 2024, ProPublica found. Washington Secretary of State Steve Hobbs, a Democrat, said he saw the service changes directed by Republican appointees leading the Postal Service as a way to undermine mail-in voting. “I don’t think they realized how much it was going to affect their voter base,” he said, adding that Washington has “a lot of rural Republican voters.”

Many areas with spikes in late ballots were in Western states, which were the earliest and most enthusiastic adopters of expanded mail-in voting. Washington became the second state to start conducting all-mail elections in 2004, in which every eligible voter automatically receives a mail-in ballot. Most Washington residents vote by putting their ballots in drop boxes, though about 33% put them in the mail in 2024. Mail-in voting became more popular over the preceding decade as states expanded access during the COVID-19 pandemic, and nearly 1 in 3 Americans voted by mail in both 2022 and 2024. ProPublica noted that Trump frequently votes by mail and recently encouraged supporters to do so, though he claimed without evidence that the process was rife with fraud.

Postal Service leaders say such delays are less likely on Nov. 3 because of the “extraordinary measures” the agency uses for these votes, when turnout is typically far higher. A memo sent to more than 30,000 post offices says that during the final 10 days before the election, managers should separate completed ballots from other first-class mail and deliver ballots intended for local election offices on the day they arrive. At an Oct. 15 news conference, Postal Service leaders reiterated the recommendation that voters mail ballots at least a week before their state deadlines and defended the agency’s ability to fast-track ballots through the system.

Postal Service board of governors chair Amber McReynolds said, “We work to identify and address issues quickly,” adding that the goal was “so ballots can continue to move swiftly from election offices to voters and from voters back to election offices.” Election consultants, researchers and former Postal Service officials said the November guidance gave agencies more detailed directions for expediting ballots. They hoped the special measures would limit delays, though some said mistakes were likely. Tammy Patrick, chief programs officer of the National Association of Election Officials, said, “I don’t think we can say with any kind of conviction that there will be no ballots rejected,” adding, “It isn’t a question of ‘if,’ but a question of ‘how many.’”

A pilot program instituting the delivery changes in 18 regions before the 2024 election ran into problems. An audit by the Postal Service’s inspector general found that agency staff in those areas sometimes did not handle ballots correctly, mainly because instructions were not clearly communicated. A Postal Service spokesperson said “learnings from the pilot” had been applied as the program expanded. After the 2026 primary, media outlets reported that mail-in ballot rejection rates were higher than in the past, but ProPublica’s analysis showed the increases were directly connected to Postal Service changes. Research published in September by the Brookings Institution also found this connection and projected that tens of thousands of mail-in ballots could be rejected if the pattern held in the general election.

Elena Patel, who conducted the Brookings analysis, said service changes may still imperil mail-in votes in the general election. “I think there remain risks to ballots,” she said, adding that the Postal Service’s plan “depends on flawless execution.” At least seven closely contested congressional districts—including Washington’s 3rd, California’s 22nd and North Carolina’s 11th—include areas affected during the primary.

The Postal Service has mostly ensured the timely delivery of hundreds of millions of ballots, even as overall mail volume dropped 50% from its high in 2006. When Louis DeJoy became postmaster general in spring 2020, he inherited an $81 billion operating deficit and rapidly took steps such as eliminating overtime to reduce costs under pressure from Trump and others in his administration. DeJoy did not respond to questions from ProPublica.

The NAACP sued that August, alleging the cost-cutting measures could affect mail-in votes in that year’s general election. Courts ordered DeJoy to reverse some of the cuts, and the Postal Service agreed to implement special procedures to ensure ballots were not delayed. The agency has continued following those procedures since.

DeJoy proceeded with a 10-year restructuring plan designed to save the agency billions. It ended evening pickups from some post offices, often in rural areas where trucks traveled long distances and were at times mostly empty. The portion of the plan known as RTO was projected eventually to save the Postal Service $4 billion a year.

The agency announced the national rollout of RTO in 2024, but DeJoy agreed to pause it leading up to that year’s November election, leaving about 4,000 ZIP codes in the pilot project affected by the service changes. The rollout later resumed, with changes made in five times as many ZIP codes for the 2026 midterms and almost 20,000 ZIP codes to date. Officials expected the system to be fully implemented within two years.

In January 2025, the Postal Regulatory Commission, which oversees the Postal Service, warned that advancing the plan would disrupt rural communities and might not save as much money as expected. The Postal Service proceeded. A January statement acknowledged that the transportation cuts would increase the time needed to postmark and deliver mail, including ballots, but the agency did not specify which voters would experience the slowdowns.

Although RTO’s delay is often only a day, mail can take longer when deposited after the daily pickup or shortly before Sunday or a holiday. A Postal Service spokesperson said the agency told election officials that RTO would change schedules and reiterated the recommendation to send ballots through the mail at least a week before state deadlines. Washington officials said the Postal Service had not been direct enough about the fact that the changes would lead to delays in delivering and postmarking ballots.

Hobbs said, “What they said was, ‘We are going to change the way we do postmarks,’” adding, “That was it.” Patel, a Brookings analyst and former Postal Regulatory Commission employee, said more direct communication to voters could have helped change their behavior. “That’s the part that I think has not been super transparent about the whole process,” she said. Stuart Holmes, Washington’s director of elections, said he reached out to Postal Service officials but received few answers and no specifics about RTO delays, which he called a “total surprise.”

Election officials in California and Oregon also reported thousands more ballots rejected for arriving late during their primaries. Counties affected by RTO, including Kern County, California, had substantially higher rates of ballots rejected for mail-related reasons, ProPublica found. Kern County assistant registrar of voters Laura Cantu said election officials were urging rural voters to use drop boxes in the general election.

In North Carolina and Ohio, late-ballot rates were about twice as high in RTO-affected counties as in counties that were not affected. Since 2022, both states had eliminated grace periods in which ballots postmarked by Election Day could count even if they arrived later. Voters unfamiliar with the tighter deadlines also could have contributed to the increase in late ballots.

ProPublica found no association between RTO and higher late-ballot rates in Wisconsin, even though clerks reported delays involving mail-in ballots cast in the state election held that spring; other Postal Service changes may have contributed. Wisconsin’s data does not distinguish ballots sent by mail from those cast in drop boxes or in person, making it impossible to definitively measure rejection rates for mailed ballots alone. Election researchers said the state’s deadline, requiring mailed ballots to reach election offices by 8 p.m. on Election Day, probably also had an effect. “If you know that you don’t have a grace period, it probably affects voter behavior,” Patel said.

Recent Postal Service audits found that RTO-related changes had saved the agency about $2.8 billion by the end of fiscal year 2025, less money than expected, while slowing all first-class mail rather than only ballots. Rural residents were waiting longer for prescriptions, bills and other mail, a September report from the Postal Service’s inspector general said. A Postal Service spokesperson defended the plan for financial reasons and to adjust to the changing marketplace, saying, “These savings are both significant and necessary.”

In the lead-up to Nov. 3, the agency plans to route ballots meant for local election offices to a nearby postal facility or directly to their destination rather than regional plants, which the Postal Service says should minimize delays. Larsen, the Stevens County auditor, is encouraging Washington voters to deposit their ballots in drop boxes rather than relying on the mail. Since the primary, she has increased the number of red, white and blue drop boxes outside government buildings from seven to 12. Near each box, a large sign asks voters to take ballot selfies and send the photos to Larsen. In prior years, Larsen said, she avoided telling voters how or when to vote because she worried they might regret choices if critical information later emerged about the candidates; she said that concern was no longer uppermost. “Now we are more concerned that we get your ballot,” she said.

To quantify the effects, ProPublica compiled affected ZIP codes and compared them with ballot-rejection tallies from nine states. Only Washington, North Carolina and Oregon tracked rejections with enough detail to distinguish mailed ballots from those received by drop box or another method. In those three states, the analysis found a direct relationship between counties where RTO had been implemented and increased rejection rates for mailed ballots received late in the 2026 primaries. A ballot was considered late if it arrived after the permitted deadline or carried a postmark after the deadline.

In California and Ohio, where data were less detailed, ProPublica connected the logistics changes with higher rejection rates generally, which experts said were likely tied to Postal Service changes. California was missing 2022 primary data for four counties covering 2% of registered voters. ProPublica could not establish an effect in Colorado, Pennsylvania and Wisconsin because those states do not accept ballots after Election Day.

The Postal Service does not publicly maintain a master list of delivery ZIP codes affected by RTO. ProPublica used filings in the Postal Regulatory Commission’s online docket, which listed the addresses of individual postal facilities eligible for RTO. It treated a ZIP code as affected if at least one post office on the lists served it. When filings omitted a post office’s delivery ZIP codes, ProPublica assumed the ZIP code in the office’s physical address had undergone RTO, and it used the earliest implementation date when a ZIP code had more than one. ProPublica also published a lookup tool showing whether RTO had been implemented in a ZIP code.

Because the Postal Service does not maintain official ZIP-code boundaries, ProPublica used ZIP Code Tabulation Areas from the U.S. Census Bureau. A county was considered affected when a majority of its voting-age citizens lived in ZIP codes where RTO had been implemented at the time of the state primary. Population data came from the American Community Survey five-year special tabulation for 2024, the most recent year available. The results did not change after accounting for margin of error. ProPublica interpolated block-group data to the ZIP Code Tabulation Areas and assumed the population was evenly distributed. The analysis credited graphics by Chris Alcantara and Ken B. Morales, with Ruth Talbot and Ellis Simani contributing data reporting.

ProPublica’s clearest direct finding is in the three states with granular mail-ballot tracking. Elsewhere, state-level detail and geographic assumptions limit precision. The Postal Service says its final-10-day separation, daily local-ballot delivery and routing plan should minimize delays, while its leaders say such delays are less likely on Nov. 3. Outside election specialists said they hoped the safeguards would limit delays, but some said mistakes were likely, and Patel said the plan depends on flawless execution.