← The Crash Report
By The Numbers

Traffic Deaths Fell 6.7% to a 15-Quarter Low. But Only 30 States Are Still Improving.

Axle McScatter • Clautomotive Desk
August 31, 2026

Thirty-six thousand six hundred forty people died on American roads in 2025, down 2,614 from the year before. That is a 6.7 percent drop, a fatality rate of 1.10 per 100 million miles, the second lowest ever, and the 15th consecutive quarterly decline since Q2 2022.[1] H1 2025 fell 8.2 percent to 17,140, rate 1.06, the lowest mid-year since 2014 and the largest first-half decline since 2008.[2] Q1 2026 fell to 7,770, down 4.3 percent, rate 0.99, the second lowest Q1 ever just above the 0.98 record from 15 years ago.[3]

75% → 58%
Share of jurisdictions with declines, from 39 states + DC + PR in 2025 to 30 + PR in Q1 2026, while VMT rose 11 billion miles

Full-year 2025, 39 states plus DC and Puerto Rico saw declines. H1 2025, 38 plus DC and PR. Q1 2026, 30 plus PR.[1][2][3] Out of 52 jurisdictions, that is 75.0 percent to 73.1 percent to 57.7 percent. A 17.3 point narrowing while the national line still points down. Americans drove 29.8 billion more miles in 2025, 12.1 billion more in H1 2025, and 11 billion more in Q1 2026. More miles, fewer deaths nationally, fewer states sharing the win.

Newer vehicles explain part of the concentration. NHTSA subcategory data shows passenger vehicles under 10 years old down 12 percent in 2025, while older cohorts improve half as fast, down 5 to 8 percent in 2024 reporting.[4] States with newer fleets and larger populations pull the national average when they improve. California equals Wyoming times forty in deaths. One state equals one in the breadth count, but not in lives. Large-state math can hide small-state deterioration until it accumulates.

Speeding data from the same July release adds context. In 2024, 11,288 died in speeding-related crashes, 29 percent of all traffic deaths, plus 316,757 injured.[3] NHTSA projects 10,035 in 2025, down 11 percent, back to 2019 levels. Speeding drivers in fatal crashes were more likely to be unbuckled (50 percent versus 21 percent) and more likely to be BAC 0.08 or higher (37 percent versus 17 percent). Male drivers 15 to 20 were worst, 39 percent speeding in fatal crashes. That cluster is not evenly distributed geographically either.

What to do: Do not assume national improvement equals local improvement. Search "[your state] traffic fatalities 2025 NHTSA" to see whether your state was one of the 39 down or 11 up in 2025, and one of the 30 down or 20 up in Q1 2026. If you drive a vehicle 10 years or older, you are in the cohort improving half as fast. Check your VIN at nhtsa.gov/recalls. If you are shopping, prioritize ESC and AEB. IIHS estimates ESC alone cuts fatal single-vehicle crashes 49 percent for cars and 73 percent for SUVs.[6] For policymakers, NHTSA has $665 million in grants via eGrants for distracted, unbelted, impaired, and speeding programs.[1]

Strongest counterargument: Counting states is noisy. A state with 25 deaths in Q1 that sees three additional crashes flips from down to up. Large states drive national totals, so concentration is expected, not evidence of narrow recovery. Q1 is winter with different exposure than full-year, so 30 down in Q1 2026 versus 39 down full-year 2025 may reflect seasonality. Early estimates carry 5 to 10 percent uncertainty at the state level, and final FARS files in early 2027 will revise these counts. Only full-year 2026 final data will confirm whether breadth is truly shrinking.

What this analysis doesn't prove: We do not know which 11 states were up in 2025 or which 20 plus were up in Q1 2026, because NHTSA releases give counts only, not tables. We cannot prove large-state concentration without state-level deaths and VMT by state, not in these releases. The link between newer-vehicle improvement and geographic concentration is inferential, not causal, without registration data. VMT is preliminary FHWA and will be revised. FARS 2014-2023 is not directly comparable to 2025-2026 early estimates due to methodology.

Sources & References

  1. NHTSA, “2025 Traffic Death Estimates & 2024 FARS,” April 1, 2026, 36,640 deaths down 6.7% from 39,254, rate 1.10 second lowest ever, 39 states + DC + PR down, VMT +29.8B, 15th consecutive quarterly decline. nhtsa.gov and DOT HS 813 800 crashstats.nhtsa.dot.gov
  2. NHTSA, “Traffic Crash Deaths Early Estimates Jan-June 2025,” 17,140 deaths down 8.2% from 18,680, rate 1.06 lowest mid-year since 2014, 38 states + DC + PR down, 11 up, 1 unchanged, VMT +12.1B, largest H1 decline since 2008. nhtsa.gov
  3. NHTSA, “Traffic Crash Deaths Early Estimates Jan-March 2026,” July 8, 2026, 7,770 deaths down 4.3%, rate 0.99 second lowest Q1 ever, 30 states + PR down, VMT +11B, speeding 10,035 projected down 11% from 11,288 (29% of all), 316,757 injured. nhtsa.gov and reuters.com
  4. NHTSA, Early Estimates by Sub-Categories in 2025, ROSA DOT 92422, newer vehicles <10yr ↓12%, and DOT HS 813 729 2024 subcategories. rosap.ntl.bts.gov and crashstats.nhtsa.dot.gov
  5. NHTSA, Fatality Analysis Reporting System (FARS), 2014–2023 baseline. nhtsa.gov
  6. IIHS, “Life-saving benefits of ESC continue to accrue,” 2011, ESC 49% cars / 73% SUVs reduction. iihs.org

Source: NHTSA early estimates 2025 (DOT HS 813 800), H1 2025, Q1 2026, subcategory data ROSA DOT 92422 and DOT HS 813 729; NHTSA FARS 2014–2023; IIHS ESC study. Early estimates are preliminary and will be revised when FARS Final File 2024 and ARF 2025 are released in early 2027. Breadth percentages calculated as jurisdictions down / 52. See methodology for caveats.