Society | Road safety

When the clocks go back, the walk home turns deadly

In the week after America turns its clocks back, two and a half times as many people die on foot between 5pm and 7pm as in the week before. Fifty years of crash records, and a change of rule in 2007, show that the darkness is to blame

November 2026 5 min read Data Cite this work

2.6×Pedestrians killed between 5pm and 7pm in the week after the clocks go back, against the week before. 1,869 against 710, 1975-2024 added together
35 → 14Pedestrians killed between 5pm and 7pm a year in the week that moved in 2007, before the rule change and after it
About 25Lives a year saved by moving the change a week later, after counting the extra deaths in darker mornings
46% → 88%Share of the evening's pedestrian deaths that happened in the dark, in the three weeks before the change and the three weeks after

On 1 November most of America turned its clocks back an hour. Overnight the evening rush hour, which had been in daylight or dusk, moved into the dark. America's federal register of fatal crashes, kept since 1975, records every death on a public road with its date and hour. We counted the pedestrians killed in the three weeks before each autumn change and the three weeks after, hour by hour, for fifty years.

Between 5pm and 7pm the toll jumps the moment the clocks change. In the week before, 710 people were killed on foot at those hours, all fifty years added together. In the week after, 1,869 were, two and a half times as many, and the count stays there for the following weeks. Late at night, from 8pm to midnight, when it is dark either way, deaths do not rise; they fall slightly.

3 weeks before 2 weeks before Week before Week after 2 weeks after 3 weeks after 486 642 710 1,869 1,797 1,826 Clocks go back 0 500 1,000 1,500 2,000 When the clocks go back, the evening turns deadly Pedestrians killed in the US by week around the autumn clock change, 1975-2024 added together 5pm to 7pm 8pm to midnight (dark throughout) Sources: NHTSA Fatality Analysis Reporting System; Signum Average calculations Left out: Arizona and Hawaii (no clock change); Indiana before 2006

The police record the light at each crash. In the three weeks before the change, 46% of the pedestrians killed between 5pm and 7pm died in the dark. In the three weeks after, 88% did. In the past decade alone, about 125 people a year were killed on foot at those hours in the three weeks after the change, against 38 in the three weeks before.

A test the calendar set up

Part of that jump is the season: the days shorten through October and November whatever the clocks say. Arizona and Hawaii, which do not change their clocks (apart from the Navajo Nation in Arizona), show it. Their evening deaths rise too, against late-night deaths, by 1.8 times. Elsewhere the rise is 3.6 times.

A change of law separates the two. Until 2006 the clocks went back on the last Sunday of October. Since 2007 they go back on the first Sunday of November. The week in between kept light evenings under the new rule, on the same dates of the year. Under the old rule (1987-2006) an average of 35 people a year were killed on foot between 5pm and 7pm in that week. Under the new rule (2007-2024) it was 14. In the two weeks after, in standard time under both rules, the toll barely moved: 33 a week and 36.

The week that moved (late Oct to early Nov) The two weeks after it (a week, on average) 35 33 14 36 0 10 20 30 40 A week of light evenings, a third of the deaths Pedestrians killed between 5pm and 7pm, a year on average. Old rule: clocks back on the last Sunday of October. New rule: the first Sunday of November, so the week in between keeps light evenings Old rule, 1987-2006 New rule, 2007-2024 Sources: NHTSA Fatality Analysis Reporting System; Signum Average calculations Left out: Arizona and Hawaii (no clock change); Indiana before 2006

The light was paid for in the morning. Under the new rule that week's mornings stayed dark for longer, and deaths between 6am and 8am more than doubled. But mornings kill far fewer people on foot to begin with: about 8 more deaths a year in the morning, against 33 fewer in the evening, once the rise in late-night deaths over the same years is allowed for. Net, the later change saves about 25 lives a year (our range: 16 to 34).

Both ways, every time

The pattern holds wherever the light moves. After the autumn change, mornings brighten and morning deaths fall by about two-fifths. After the spring change, evenings brighten and evening deaths fall by more than half, while darker mornings bring a third more deaths. The researchers Michael Sullivan and Michael Flannagan used the same clock changes in 2002 to measure how much more dangerous darkness is for people on foot. Twenty more years of records point the same way, and the evening rise has been larger since 2000 than before it.

¼ ½ 1 (no change) 2× 4× Autumn, 5-7pm Autumn, 6-8am Spring, 5-7pm Spring, 6-8am 2007 rule, moved week, 5-7pm 2007 rule, moved week, 6-8am Arizona and Hawaii, autumn, 5-7pm 3.64 0.62 0.45 1.32 0.30 2.15 1.83 Darker hours kill more people on foot, both ways Pedestrian deaths after the clock change against before, relative to 8pm-midnight, 1975-2024. 2007 rule: the moved week, new rule against old. Lines: 95% ranges Sources: NHTSA Fatality Analysis Reporting System; Signum Average calculations Arizona and Hawaii keep their clocks; their rise is the natural shortening of the days

Congress has argued for years about ending the change, and the Senate voted in 2022 to make daylight saving time permanent. The register cannot settle that argument: it counts deaths, not how many people walk at each hour. It does show where the danger lies now. In November to February, the hour from 6pm is the deadliest of the day for people on foot, with almost three and a half times the deaths of the worst morning hour. Until the rules change, the next few weeks are the time to wear something bright on the walk home, and to drive as if someone is out there.

Data

Pedestrian deaths by day around the autumn clock change

Pedestrians killed on each day from three weeks before the autumn clock change to three weeks after, by hour band, all years 1975-2024 added together. Coverage: 1975-2024. Source: NHTSA FARS; own calculations. clock-by-day.csv

Pedestrian deaths around the autumn clock change, year by year

For each year, the date the clocks went back and the pedestrians killed in each hour band in the 21 days before (days -21 to -1) and the 21 days from it (days 0 to 20). Coverage: 1975-2024. Source: NHTSA FARS; own calculations. clock-by-year.csv

US clock change dates, 1975-2024

The Sunday the clocks went back in autumn and forward in spring, each year, by the federal rule of the time. Coverage: 1975-2024. Source: NHTSA FARS; own calculations. clock-change-dates.csv

Light at the scene, evening pedestrian deaths around the autumn change

Pedestrians killed 17:00-18:59 in the three weeks before and after the autumn change, by the light condition the police recorded, 1975-2024. Coverage: 1975-2024. Source: NHTSA FARS; own calculations. clock-light.csv

Pedestrian deaths by hour of day, November to February

Pedestrians killed in November, December, January and February, by hour of the day, 1975-2024 added together. Coverage: 1975-2024. Source: NHTSA FARS; own calculations. clock-winter-hours.csv

Every test of the clock change

Each comparison in the study: the change in pedestrian deaths in an hour band after a clock change against before, relative to 20:00-23:59, with its 95% range and the counts behind it. Coverage: 1975-2024. Source: NHTSA FARS; own calculations. clock-tests.csv

Lives a year spared by the 2007 rule

Pedestrian deaths a year spared in the week that moved in 2007 (exploratory): expected under the old rule against observed under the new, 2007-2024, with 95% ranges from 20,000 Poisson draws. Coverage: 1975-2024. Source: NHTSA FARS; own calculations. clock-lives-2007.csv

All the files, with a README that explains every column, are in the data on GitHub, free to reuse with credit (CC BY 4.0).

Cite this work

Our articles and data visualisations rely on work from many sources. When citing this article, please also cite the underlying data sources, listed with the data.

Citation

Signum Average (2026) - "When the clocks go back, the walk home turns deadly" Published online at signumaverage.com. Retrieved from: https://signumaverage.com/articles/clock-change/ [Online resource]

BibTeX
@article{signumaverage-2026-11-11-clock-change,
    author = {{Signum Average}},
    title = {When the clocks go back, the walk home turns deadly},
    journal = {Signum Average},
    year = {2026},
    month = {nov},
    url = {https://signumaverage.com/articles/clock-change/},
    note = {Online resource}
}

Reuse this work freely

Text, charts and data on this page are published under the CC BY 4.0 licence: you may copy, adapt and republish them for any purpose, as long as you credit Signum Average. Third-party data keeps its original licence.