Afternoon rush hours generate the most crash reports; late-night hours carry up to 1.76x the severe share
Across 74,533 reported California collisions in the 90-day Pulse window, afternoon commuter hours (15:00 to 17:00) produce over 5,300 crashes per hour at below-average severity (0.77x to 0.87x), whereas every hour from 20:00 through 03:00 exceeds 1.40x the statewide severe-injury baseline.
- Observation window
- 90-day statewide Crash Pulse window (2026-07-30 to 2026-10-27, 74,533 collisions)
- Primary records
- CHP CCRS public collision records (statewide hourly pulse aggregation)
- Editorial date
- Published 2026-11-12 (Updated 2026-11-12)
Highest hourly crash volume, with 2.88% severe share (0.83x statewide 3.48% baseline).
6.14% of crashes between 10:00 PM and 10:59 PM were fatal or severe (133 of 2,167, 9 killed).
5.79% severe share (62 of 1,071 crashes, 5 killed) on uncongested early-morning roadways.
2,597 fatal or severe-injury crashes and 160 people killed across 74,533 statewide collisions.
Why volume-only heatmaps misallocate high-severity countermeasures
When transportation agencies or researchers plot all reported collisions without stratifying by time of day and injury severity, afternoon commuter congestion dominates every visual summary. Between 2026-07-30 and 2026-10-27, the three afternoon hours of 15:00 (5,449 crashes), 16:00 (5,360 crashes), and 17:00 (5,321 crashes) accounted for 16,130 collisions, or 21.6% of all statewide crash filings.
Yet congestion physically constrains impact speed. During those three afternoon peak hours, fatal or severe injury share remained well below the 3.48% statewide baseline: 2.88% at 15:00 (0.83x lift), 2.69% at 16:00 (0.77x lift), and 3.03% at 17:00 (0.87x lift), totaling 24 deaths across 16,130 collisions.
Once evening traffic thins and free-flow speeds return on both freeways and multi-lane arterials, the severity profile inverts. Across the four hours from 20:00 through 23:59 and continuing through 03:00, collision volume falls sharply while severe-injury share climbs to between 4.89% and 6.14%, peaking at 22:00 (1.76x baseline) and 03:00 (1.66x baseline).
| Hour of day (local) | Reported crashes | Severe crashes | Severe share | Severity lift vs 3.48% | People killed |
|---|---|---|---|---|---|
| 02:00 to 02:59 | 1,308 | 68 | 5.20% | 1.49x | 8 |
| 03:00 to 03:59 | 1,071 | 62 | 5.79% | 1.66x | 5 |
| 08:00 to 08:59 (AM commute) | 3,842 | 98 | 2.55% | 0.73x | 5 |
| 15:00 to 15:59 (PM volume peak) | 5,449 | 157 | 2.88% | 0.83x | 7 |
| 16:00 to 16:59 | 5,360 | 144 | 2.69% | 0.77x | 8 |
| 17:00 to 17:59 | 5,321 | 161 | 3.03% | 0.87x | 9 |
| 20:00 to 20:59 | 3,191 | 156 | 4.89% | 1.40x | 21 |
| 21:00 to 21:59 | 2,681 | 142 | 5.30% | 1.52x | 10 |
| 22:00 to 22:59 (Peak lift) | 2,167 | 133 | 6.14% | 1.76x | 9 |
Matching engineering and operational interventions to the right clock window
The divergence between afternoon crash volume and late-night crash severity carries direct implications for corridor design, signal timing, and enforcement scheduling. Afternoon congestion clusters respond to queue-warning systems, ramp-metering coordination, and Changeable Message Sign (CMS) shockwave damping that reduce secondary rear-end collisions when freeway speeds drop abruptly.
Late-night severe crashes occur under fundamentally different physical conditions: wide lanes designed for peak-hour throughput invite excessive free-flow speeds on empty roadways, impairment and fatigue incidence rise, darkness reduces pedestrian and fixed-object conspicuity, and emergency trauma discovery can take longer on rural or low-volume segments.
Analysts reviewing corridor candidates should evaluate daytime congestion crashes and 20:00-to-03:00 severe collisions as two separate problem statements rather than blending 24-hour counts into a single unweighted priority score.
- Use severity-weighted or time-windowed filters on the California Crash Map to isolate corridors where 20:00 to 03:00 severe crashes cluster repeatedly.
- Avoid resting signal-timing or road-diet evaluations solely on total crash counts; convert high-speed off-peak geometry (curb extensions, raised medians, night-time speed-table or signal progression adjustments) around severe-injury hours.
- Check reporting-lag bias on recent 30-day snapshots: in the latest 30-day map window (2026-09-28 to 2026-10-27, 15,701 crashes, 444 serious-or-fatal injury records, 19 killed), late-arriving fatal investigations can still amend recent weeks.
Methodology boundaries and exposure limitations
- Hourly buckets use the officer-recorded collision timestamp in local California time; unrecorded or midnight-defaulted timestamps in raw filings are excluded from hourly lift ratios.
- Property-damage-only collisions are more likely to go unreported late at night or involve hit-run departures, which can modestly raise the observed severe percentage relative to daytime commuter hours.
- Hourly Vehicle Miles Traveled (VMT) denominators are not available statewide across surface streets, so severity lift measures severe crashes per reported collision rather than severe crashes per mile driven.
Review full source, coordinate-derivation, and report-revision rules on our Sources, Methodology, and Privacy page.
How to cite this brief
Evidence for Zero (Meru AI), "Late-Night Crash Severity Lift & Time Windows," built from CHP CCRS public collision records (90-day Pulse window 2026-07-30 to 2026-10-27), https://crashes.meruai.co/insights/late-night-severity-lift-and-time-of-day-windows.
Inspect the underlying corridors and records
Inspect all 24 hourly severity bars, ZCTA outliers, and road-pair clusters.
Slice California collision records by date window, severity, and place.
See how daytime congestion waves form around freeway incidents.
Buffer reported collisions along any California commute or freight route.