Fact
Fatal car crashes rise 6% in the week after clocks spring forward
In US crash data from 1996–2017, the risk of a fatal crash on the five workdays after the spring switch to Daylight Saving Time was 6% higher than in other weeks of the year (IRR 1.06, 95% CI 1.03–1.09). That is about 28 extra fatal crashes a year, or 627 over 22 years. The fall switch back showed no effect.
Confidence: Single studyOne study so far. Treat it as a strong lead, not a law.Show the data as a table
| label | Fatal crash rate vs. any other week (1.00 = no change) |
|---|---|
| Week before DST | 1 IRR |
| DST week | 1.06 IRR |
| Week after DST | 1.02 IRR |
Results text, "Spring DST Acutely Increases Fatal MVA Risk": DST week IRR 1.06 (95% CI 1.03–1.09); week before 1.00 (0.98–1.03); week after 1.02 (0.99–1.04). All are Monday–Friday incidence rate ratios versus any other Monday–Friday of the year.
The paper counts fatal crashes, not deaths. "About 28" is the paper's 28.5 extra fatal crashes per spring-forward workweek (95% CI 15.4–41.6), or 626.9 of 8,958 such crashes over 1996–2017. The number of deaths is not reported. The rise was largest before noon (IRR 1.09) and in the western parts of time zones, where mornings stay darker. That pattern fits sleep loss and circadian misalignment better than changes in daylight. The analysis has no data on traffic volume, weather, alcohol or how much drivers had slept, so the link to sleep is inferred. Earlier studies of the clock change were inconsistent: a 2017 systematic review of 24 studies found short-term results that went both ways. Fritz et al. is the largest and most detailed analysis so far, but it is one analysis of one country's data.
Source and method
- Fritz J, VoPham T, Wright KP Jr, Vetter C (2020). A Chronobiological Evaluation of the Acute Effects of Daylight Saving Time on Traffic Accident Risk. Current BiologySample: 732,835 fatal motor vehicle crashes, contiguous US (excluding Arizona and Indiana), 1996–2017
- Carey RN, Sarma KM (2017). Impact of daylight saving time on road traffic collision risk: a systematic review. BMJ OpenSample: 24 studies
The study used NHTSA's Fatality Analysis Reporting System (FARS), which records every US crash on a public road in which someone dies within 30 days. Alaska, Hawaii, the territories, Arizona and Indiana were excluded (the last two did not fully observe DST for most of the period). Daily crash counts were modelled with Poisson regression, adjusting for year, day of week and season (B-splines). Monday–Friday of the DST week was compared with any other Monday–Friday of the year, with the weeks before and after as controls. The 2007 change in DST dates (from early April to mid-March) served as a natural check: the spike moved with the transition date.
Published October 14, 2026.
Related facts
- FactYour body clock runs latest at about 20In ~25,000 people, the midpoint of sleep on free days drifted later through adolescence and peaked at about age 20 (19.5 for women, 20.9 for men) before moving steadily earlier for the rest of life.Confidence: Solid
- FactThere's no clear evidence that adults sleep less than 50 years agoA meta-analysis of 168 studies (257 data points, 6,052 healthy adults aged 18–88) that measured sleep objectively between 1960 and 2013 found no significant link between the year of a study and how long people slept. Reviews of self-reported sleep are mixed, though, and some large surveys do show a decline.Confidence: Debated
- FactPeople living without electricity don't sleep more than we doIn three pre-industrial societies with no electricity (the Hadza of Tanzania, the San of Namibia and the Tsimane of Bolivia), wrist actigraphy put average sleep at 5.7 to 7.1 hours a night across 10 recording groups (6.4 h on average), with sleep periods of 6.9 to 8.5 hours.Confidence: Debated