Night vs Day: When Do Self-Driving Cars Crash Most?

If you assumed self-driving cars crash most at night — when visibility is poor and sensors are challenged — you'd be wrong. The NHTSA data tells a different story: 4 PM is the deadliest hour, with 343 incidents. The safest? 4 AM, with the fewest crashes in the entire dataset. The pattern follows human traffic, not machine capability.

The Time-of-Day Curve

The incident distribution follows a bell curve centered on the late afternoon. Incidents begin climbing around 7 AM with the morning commute, dip slightly at midday, then surge to the daily peak between 3 PM and 6 PM. After 8 PM, the count drops sharply and stays low through the early morning hours.

This pattern almost perfectly mirrors overall U.S. traffic volume. More cars on the road means more interactions, more conflict points, and more crashes — whether the driver is human or silicon.

Why Afternoon Is Worst

Several factors converge during the 3–6 PM window:

• Peak traffic density — more vehicles in closer proximity increases rear-end and intersection crash probability
• Sun glare — low-angle afternoon sun is a known problem for camera-based systems, causing temporary blindness and false detections
• Driver fatigue — for ADAS vehicles requiring human supervision, end-of-day fatigue reduces takeover response times
• Complex scenarios — school zones, pedestrians, cyclists, and delivery vehicles all peak in the afternoon

The Nighttime Surprise

Nighttime hours (10 PM – 5 AM) account for a relatively small share of AV incidents. This is partly traffic volume, but it may also reflect a genuine sensor advantage: lidar and radar — used by Waymo, Cruise, and other ADS operators — perform equally well in darkness. For camera-reliant systems like Tesla, headlights and street lighting provide sufficient illumination for most scenarios.

That said, the nighttime incidents that do occur tend to be more severe. Higher average speeds and the presence of impaired human drivers (who may collide with AVs) push the severity up even as the count stays low.

ADAS vs ADS: Different Patterns

When you separate ADAS (Tesla, Honda, etc.) from ADS (Waymo, Cruise), the curves diverge. ADAS incidents closely track commuter traffic because ADAS vehicles are consumer cars used for commuting. ADS incidents are more spread across the day because robotaxi services operate on demand schedules — Waymo vehicles run from early morning through late night in cities like San Francisco and Phoenix.

What This Means