What the crash data shows
- Walking · 2021–2026 (2026 so far)
People hit on foot in the dark with no street lights are killed about four times as often
Of pedestrian crashes in the dark on streets with no lights, 29% killed the person on foot (1,570 of 5,370). In every other lighting it was 6.6% (4,014 of 61,085).
Share of pedestrian crashes that killed the person on foot - Dark, no street lights
- Other lighting
Found in 2021–2023 (32% vs 7.4%); held in 2024–2026 (26% vs 5.7%).
- Walking · 2021–2026 (2026 so far)
A crash with a person on foot on a freeway kills them more than a third of the time
Of pedestrian crashes on freeways, 37% were fatal (1,368 of 3,743), against 6.9% off the freeway (4,370 of 63,450).
Share of pedestrian crashes that killed the person on foot - On a freeway
- Off the freeway
Found in 2021–2023 (39% vs 7.7%); held in 2024–2026 (34% vs 6%).
- Walking · 2021–2026 (2026 so far)
On roads posted 50 mph or more, a third of pedestrian crashes are fatal
Where the posted limit was 50 mph or more, 33% of pedestrian crashes killed the person on foot (1,884 of 5,705). On all slower roads together it was 6.3% (1,616 of 25,666).
Share of pedestrian crashes that killed the person on foot - Posted 50 mph or more
- Slower roads
Found in 2021–2023 (35% vs 7.3%); held in 2024–2026 (30% vs 5.5%).
- Walking · 2021–2026 (2026 so far)
Even in a crosswalk at an intersection, crashes at night on a lit street are about three times as deadly
Among pedestrian crashes in a crosswalk at an intersection, 6% of those in the dark on a street with lights were fatal (517 of 8,622), against 2.1% in other light (373 of 17,729).
Crosswalk crashes at intersections: share fatal - Dark, street lit
- Other light
Found in 2021–2023 (6.8% vs 2.3%); held in 2024–2026 (5.2% vs 1.9%).
- Walking · 2021–2026 (2026 so far)
When the clocks go back in November, pedestrian crashes move into the dark
In the four weeks after the November clock change there were 2,887 pedestrian crashes in the dark, against 2,436 in the four weeks before: 19% more. Crashes in daylight went the other way: 2,969 before, 2,426 after.
Pedestrian crashes in the dark - 4 weeks before
- 4 weeks after
Found in 2021–2023 (1,751 vs 1,451); held in 2024–2026 (1,136 vs 985).
- Cycling · 2021–2026 (2026 so far)
Bike crashes in the dark with no street lights are about seven times as often fatal
Of bike crashes in the dark on unlit roads, 9.5% killed the rider (138 of 1,456), against 1.3% in all other lighting (718 of 55,543).
Share of bike crashes that killed the rider - Dark, no street lights
- Other lighting
Found in 2021–2023 (11% vs 1.5%); held in 2024–2026 (7.7% vs 1.1%).
- Cycling · 2021–2026 (2026 so far)
Rear-end crashes are a small share of bike crashes but about five times as often fatal
5.2% of bike crashes classed as rear-end collisions were fatal (180 of 3,486), against 1.1% of all other bike crashes (417 of 38,056).
Share of bike crashes that killed the rider - Rear-end
- Other bike crashes
Found in 2021–2023 (6% vs 1.2%); held in 2024–2026 (4.4% vs 1%).
- Motorcycling · 2021–2026 (2026 so far)
On roads posted 50 mph or more, head-on motorcycle crashes kill about one rider in four
On roads posted 50 mph or more, 23% of head-on motorcycle crashes were fatal (118 of 524), against 3.9% of the other crashes on those roads (1,215 of 31,409).
Roads posted 50 mph or more: share of crashes fatal - Head-on
- Other crashes
Found in 2021–2023 (23% vs 4.5%); held in 2024–2026 (22% vs 3.2%).
- Motorcycling · 2021–2026 (2026 so far)
When a motorcycle crashes alone, hitting a fixed object is about three times as deadly
Among motorcycle crashes with no other vehicle, 8.4% of those that hit a fixed object (a barrier, pole, tree or wall) were fatal (622 of 7,399), against 2.4% of the rest (285 of 11,952). Most of the rest are classed as overturned: the bike went down on the road.
Solo motorcycle crashes: share fatal - Hit a fixed object
- Other solo crashes
Found in 2021–2023 (9.5% vs 2.8%); held in 2024–2026 (7.1% vs 1.9%).
- Walking · 2021–2026 (2026 so far)
Outside city limits, a quarter of pedestrian crashes happen in the dark with no street lights
Outside city limits, 24% of pedestrian crashes were in the dark on unlit roads (2,324 of 9,887). Inside cities it was 5.4% (3,046 of 56,568).
Share of pedestrian crashes in the dark with no street lights - Outside city limits
- Inside cities
Found in 2021–2023 (26% vs 5.7%); held in 2024–2026 (21% vs 5.1%).
- Motorcycling · 2021–2026 (2026 so far)
Solo motorcycle crashes in bends happen going downhill nearly twice as often as going uphill
Outside city limits, of the single-vehicle motorcycle crashes in bends on a slope, 871 were going downhill and 483 uphill: 64% downhill.
Solo crashes in bends on a slope - Going downhill
- Going uphill
From the bend study: totals across all years, not split.
How these were found
A script tests the crash records slice by slice: how often a kind of crash kills, how one place's crashes differ from everywhere else's, and what changes around the November clock change. The last run, on 2026-10-09, tried 2,525 patterns. 362 passed a false-discovery test (so a scan of thousands doesn't “find” noise) and a minimum size, and 292 held in years the scan never looked at.
Before trusting it, the scan had to find 4 patterns already known (it found 4), such as pedestrian crashes moving into the dark after the clocks go back. Then a person read the survivors, dropped those that come from how police record crashes rather than from the streets, and kept the 11 below. Each number is recomputed from the data whenever it is rebuilt.
None of this says a street or a mode is safe or dangerous: we have no counts of how many people walk, ride or drive where. These are shares within crashes, and comparisons within the same records. Methodology