NHTSA: U.S. Traffic Deaths Fell in Early 2026, but Second-Quarter Progress Nearly Stalled
NHTSA’s new estimate shows 425 fewer deaths in the first half of 2026, but nearly all of that improvement came in the first quarter.
The United States recorded an estimated 16,750 traffic deaths in the first six months of 2026, down 2.5% from the comparable 2025 estimate, the National Highway Traffic Safety Administration reported on November 10. That is a meaningful reduction in a toll measured in human lives. It is also a national average with an important warning inside it: the decline was much stronger at the beginning of the year than at the end of the period covered by the release.
NHTSA estimates that 7,715 people died in January through March, 405 fewer than in the same quarter of 2025. From April through June, the estimated toll was 9,035, only 20 below the previous year. Together, those changes produce the half-year reduction of 425. The first quarter therefore accounts for roughly 95% of the improvement, calculated from the agency’s published estimates. May and June each moved in the opposite direction, with more deaths than in the corresponding month of 2025.
This is a November publication about January through June, not a live reading of November roads. The figures are provisional projections, not final case-by-case totals. Still, the split between the two quarters changes how the good news should be understood. America’s fatality toll continued to decline overall, and the rate per mile also improved. The latest months in this particular release do not show the same strength as the half-year headline.
A better national result, with a much narrower margin
The half-year comparison is 16,750 estimated deaths against 17,175 in the first half of 2025. NHTSA’s preliminary travel data, drawn from the Federal Highway Administration, indicate that vehicles traveled about 12.9 billion more miles, an increase of 0.8%. Fewer deaths alongside more travel is a more encouraging combination than fewer deaths caused solely by people driving less. The exposure-adjusted fatality rate fell from 1.05 to 1.02 deaths per 100 million vehicle miles traveled.
Those measures answer different questions. The count describes the size of the human toll. The rate relates that toll to the amount of vehicle travel. Neither replaces the other. A country can reduce risk per mile while still losing many people because total travel is large; it can also experience a lower count during a period when exposure falls. In this report, both the count and the rate move down while travel rises. That combination is the strongest part of the national result.
The historical comparison needs equal care. NHTSA says that, if realized, 1.02 would be the second-lowest first-half fatality rate in the comparison, behind 1.01 in the first half of 2014. That is not a claim that the number of deaths is at its lowest level. The report lists 15,035 first-half deaths in 2014, below the 16,750 now projected for 2026. Nor is 1.02 a completed full-year rate. Comparing a half-year estimate with a full-year result would mix different periods and seasonal patterns.
The human significance of 425 fewer deaths should not be lost in those distinctions. But precision is especially useful when an encouraging statistic becomes a slogan. “Lower risk per mile,” “fewer deaths than last year” and “the safest year” are not interchangeable statements. The new release supports the first two for the first half of 2026. It does not settle the third, and it does not describe how every road user or locality experienced the period.
The quarterly split is the story behind the headline
The first-quarter decline was 5.0%, from 8,120 to 7,715. The second-quarter decline was 0.2%, from 9,055 to 9,035. These are year-over-year comparisons, so the second-quarter result is being compared with the same three months of 2025, not with winter 2026. That is the appropriate starting point for a seasonal activity such as travel. Simply noting that more people died in April through June than in January through March would not isolate a change in safety.
The rate tells a similar story. In the first quarter, estimated deaths per 100 million miles fell from 1.05 to 0.99. In the second quarter, the published rate was 1.05 in both years after rounding. Travel increased 1.4% in the first quarter and 0.2% in the second. The second-quarter count was therefore only slightly lower despite essentially similar travel exposure. The broad half-year improvement is real in the estimates, but it was not evenly distributed across time.
NHTSA’s historical chart shows 17 consecutive quarters of year-over-year declines from the second quarter of 2022 through the second quarter of 2026. The latest quarter extends that sequence. Yet a sequence is not a measure of the size of each improvement. A decline of 20 deaths and a much larger decline both keep a streak intact. The new report is a reason to examine the margin, not simply celebrate the number of quarters for which the direction has remained favorable.
It would be premature to declare a reversal on the strength of one nearly flat quarter. It would be equally misleading to present the latest quarter as though the pace of improvement were unchanged. Both judgments require more information than a streak alone supplies. The useful conclusion is narrower: by the end of June, the annual comparison had become much less favorable than it was at the start of the year. The next release will help show whether that weakening persisted.
May and June push against the six-month trend
The monthly estimates make the slowdown visible. January fell from 2,690 deaths to 2,465, February from 2,465 to 2,355, and March from 2,965 to 2,895. April was also lower, at 2,760 compared with 2,915. Those four months supplied reductions of 225, 110, 70 and 155 respectively. Their combined decline was 560 deaths before the last two months of the reporting period moved the other way.
May’s estimated toll rose from 3,090 to 3,135, an increase of 45, or 1.5%. June rose from 3,050 to 3,140, an increase of 90, or 3.0%. Together those increases offset 135 of the reduction recorded in the first four months. This arithmetic is a description of the published estimates, not a causal explanation for the change. It shows why a six-month total can conceal a different pattern in the most recent part of the window.
Exposure does not provide a simple explanation for May. NHTSA’s table shows vehicle miles traveled declining 0.7% that month while deaths increased. Its estimated fatality rate rose from 1.05 to 1.07 per 100 million miles. In June, travel increased 0.9%, but deaths rose faster, and the rate increased from 1.06 to 1.09. Those comparisons make it harder to read the late-period deterioration as merely more people being on the road.
The report does not establish why those months changed. It does not provide a monthly causal decomposition into speed, impairment, distraction, vehicle design, roadway conditions or emergency response. Weather and travel composition also cannot be assigned a role from this table alone. Identifying a pattern is the beginning of an investigation, not permission to attach a preferred explanation to it. A claim about a particular cause would need matching evidence about that cause, location and period.
A national decline did not reach every state
NHTSA projects decreases in 35 states and Puerto Rico, while 15 states and the District of Columbia are projected to have increases. Puerto Rico is shown separately and is not included in the U.S. total of 16,750. Eight of the agency’s ten administrative regions had projected declines in both deaths and the fatality rate. These patterns indicate that the improvement was geographically broad, but not universal.
The largest states illustrate why the national average cannot stand in for a local experience. California’s estimate rose from 1,792 to 1,825, up 1.8%, while its rate edged from 1.08 to 1.09. Texas was almost unchanged, falling from 1,856 to 1,851, or 0.3%, with its rate moving from 1.20 to 1.19. Florida fell from 1,474 to 1,439, a 2.4% reduction, while its rate declined from 1.15 to 1.12. These are different trajectories inside one national result.
Other comparisons are larger. North Carolina’s estimated toll increased from 696 to 746, or 7.2%, and its rate rose from 1.07 to 1.14. Illinois fell from 569 to 453, a 20.4% decline, with its rate moving from 1.11 to 0.88. Those figures identify places where further investigation could be valuable. They do not, by themselves, identify which state policy succeeded or failed. Different travel patterns, road systems and reporting dynamics can complicate a direct comparison.
Small numbers require special restraint. The District of Columbia estimate rose from seven to 16 deaths, which produces a 128.6% increase. Reporting only that percentage would conceal the scale and volatility of the underlying count. Reporting only the count would conceal the seriousness of the change for the people affected. The responsible presentation includes both, identifies the estimate as provisional and avoids treating a small-base percentage as a national league table of danger.
These estimates are designed to arrive before the final census
NHTSA’s Fatality Analysis Reporting System, or FARS, is a census of fatal traffic crashes. To be included, a crash must involve a motor vehicle traveling on a trafficway and result in the death of at least one person within 30 days. The person may be a vehicle occupant or a nonoccupant. That definition is broader than “drivers killed” and different from a count of all crashes or all injuries. The 16,750 estimate should not be relabeled as any of those other measures.
Final crash data take time to assemble. This early report combines FARS history, Early Notification data, Monthly Fatality Counts and FHWA travel estimates. The agency uses statistical procedures to model relationships among the sources and produce regional and national projections. The Early Notification and Monthly Fatality Counts inputs are not publicly available in the report. Readers can inspect the published totals and methods, but cannot independently reconstruct every component from the six-page release alone.
State figures involve another step. NHTSA allocates projected regional monthly fatalities to states using the most recent relative distribution in the Early Notification data. The report describes an assumption that the inflation rate applied to states within a region is the same as that region’s rate. That technical detail matters because state estimates are not simply final, independently completed counts placed side by side. The methodology supplies timely information while leaving room for later revision.
The agency says both the fatality counts and the resulting rates and percentage changes will be refined as later FARS files and travel estimates become available. It expects the first-nine-month projections in late December. That next release will add July through September, rather than turn the present report into a full-year census. A small movement such as the second quarter’s 0.2% decline is particularly important to keep labeled as an estimate rather than an immutable final result.
What the numbers can ask of road-safety policy
A national aggregate is useful for judging the scale and direction of a problem. It is less useful for diagnosing the exact countermeasure needed at a particular crossing, curve or intersection. The National Park Service’s transportation-safety program, for example, describes safety as work spanning planning, design, construction, operation and maintenance. That is established program context, not evidence explaining the new national decline. It illustrates how many different decisions sit underneath the single outcome called a traffic fatality.
For a concrete example of how road design changes driving decisions, our guide to U.S. roundabout approaches and lane choices examines one specific intersection system. It is practical context, not an explanation for the national change reported here.
The archival photographs accompanying this article show some of those distinct settings: road maintenance, pedestrian infrastructure, enforcement equipment, restraint checks and rider training. They do not depict crashes counted in the 2026 estimate, and they do not demonstrate that any photographed activity caused the decline. Their purpose is to make the range of systems visible without substituting a dramatic crash image for evidence. A training scene can show what a program looks like; measuring its effect requires a different kind of study.
For a local authority or an interested resident, the next question is therefore where the aggregate points toward more detailed work. Are changes concentrated on particular routes or among particular kinds of road users? Do local exposure measures support the apparent trend? Has reporting completeness changed? Are the same locations repeatedly appearing in crash records? The present release does not answer those questions, but it gives a current national and state framework in which to ask them.
The next report has a clear test
The next three months of estimates will matter because they come after a quarter that was nearly flat and two months that rose. If the year-over-year reductions strengthen again, the first-half slowdown may look temporary. If they do not, a favorable full-period average could continue to obscure a weakening recent trend. Neither outcome is established today. Forecasting the full year by simply doubling 16,750 would ignore both seasonal variation and the change already visible within these six months.
The clearest reading of the November 10 release is therefore two-part. The United States had fewer estimated traffic deaths and a lower rate per mile in the first half of 2026, despite somewhat more driving. But almost all of the improvement came in the first quarter, and the final two months in the report were worse than a year earlier. The achievement deserves recognition. The thinning margin deserves attention before a continuing decline is mistaken for a problem already solved.
Comments
Post a Comment