What are the odds of causing a fatal crash by driving without enough sleep?
Evidence quality 4.5/5
Eight-dimension review score against the quality rubric . Each dimension scored 1–5.
- D1 Source grounding
- 5/5
- D2 Source authority
- 5/5
- D3 Arithmetic
- 4/5
- D4 Uncertainty
- 4/5
- D5 Scope
- 5/5
- D6 Prose
- 4/5
- D7 Perception honesty
- 4/5
- D8 Caveat completeness
- 5/5
Lifetime probability · lifetime, activity-specific
1 in 26
3.8% lifetime chance
Most people underestimate this.
range 1 in 56 to 1 in 9.1
● your factors — click this risk ▾ to reveal
≈ As likely as
Perceived
Drowsy driving is one of the most consistently underrated risks on the road, not because people deny that fatigue is dangerous, but because they trust their ability to manage it. Most drivers believe they can sense when they are too tired to drive — that the eyelid heaviness, the head nod, or a yawn will reliably tell them to pull over. Survey data and naturalistic dashcam studies show the opposite: the moments before a drowsy crash are usually not recognised by the driver, and the most common subjective state preceding a fatigue-related crash is feeling "fine, just a bit tired". A meaningful minority of drivers also believe that coffee, loud music, or rolling the window down restore alertness for more than the few minutes they actually do.
Rough estimate: most drivers believe they can tell when they are too tired to drive
Source: editorial intuition, not polled
Actual
~8 per 100,000 trips result in a fatal crash for a driver who is moderately sleep-deprived (≈4× the rested-driver rate)
US adult driver after 4–5 hours of sleep, fatal-crash involvement rate derived from Tefft 2018 culpable-crash odds applied to NHTSA per-trip baseline
Show derivation
The US population-average per-trip fatal-crash probability for a rested sober driver is approximately 1 in 50,000 (NHTSA per-trip baseline used in the driving-at-0.1pct-bac entry). Tefft 2018 (Sleep journal, peer-reviewed case-control derived from NMVCCS) found that drivers who slept 4–5 hours in the 24 hours before driving had approximately 4.3× the odds of being culpable for their crash compared with drivers who slept 7+ hours; <4 hours rose to roughly 11–15× culpability odds. Applying a conservative 4× per-trip multiplier to the rested baseline gives ~1 in 12,500 per moderately drowsy trip. For a driver who operates at this level of sleep deprivation roughly once a month — a common pattern for shift workers, new parents, and people routinely driving home from long days — 12 trips per year over 40 driving years equals ~480 impaired trips. The cumulative probability of being involved in at least one fatal crash is 1 − (1 − 1/12500)^480 ≈ 0.038, or roughly 1 in 26. The low end of the uncertainty band assumes rare drowsy trips (~6/year) with only moderate fatigue; the high end assumes monthly severe sleep deprivation (<4 hours sleep), which Tefft's 11–15× multiplier pushes toward ~1 in 9 lifetime.
Caveats: The lifetime estimate is highly sensitive to two assumptions: how often the driv…
The lifetime estimate is highly sensitive to two assumptions: how often the driver is significantly sleep-deprived, and how the per-trip multiplier from Tefft's all-severity culpable-crash analysis translates to fatal-crash risk specifically. Fatal crashes likely cluster at the more severely impaired end of the distribution, so the headline figure may understate the high-frequency severe-sleep-loss case and overstate the rare moderate case. Drowsiness is also a risk factor that operates differently from BAC: a driver at a steady 0.10% BAC is impaired for the duration of the trip, but a sleep-deprived driver's risk spikes during the late-trip window when microsleeps become more frequent. The 17.6% figure for fatal-crash prevalence (AAA 2024) and the 1.8% police-reported figure (NHTSA FARS) bracket the true population share; the order-of-magnitude gap reflects the difficulty of attributing a crash to drowsiness post-hoc when the driver may not survive to report and no objective biomarker exists. None of this accounts for the interaction with alcohol or sedating medication, which compounds impairment far beyond the additive sum of the components.
Risks at similar odds
Other risks with roughly the same likelihood — useful for calibration.
Driving at 0.10% BAC
What are the odds of causing a fatal crash when driving with a 0.10% blood alcohol level?
Driving after cannabis
What are the odds of causing a fatal crash by driving within a few hours of using cannabis?
Video watching + driving
What are the odds of a crash from watching video on a phone while driving?
Spaceflight (astronaut)
What are the odds of dying as an astronaut on a spaceflight mission?
Cruise ship norovirus
What are the odds of being on a cruise ship voyage that has a norovirus outbreak?
Driving on sedating meds
What are the odds of causing a fatal crash by driving while on a sedating prescription medication?
Recently viewed on this device
Stored locally — clear anytime.
Pick challenger
The most defensible recent estimate of the true scale of drowsy driving comes from the AAA Foundation for Traffic Safety’s 2024 reanalysis, which applied a model trained on NHTSA’s Crash Investigation Sampling System — where investigators reach detailed causation determinations — to the FARS census of fatal crashes. The result was that approximately 17.6% of fatal crashes from 2017 to 2021 involved a drowsy driver, or roughly 6,000 deaths per year (29,834 over the five-year period). That figure is about ten times higher than the NHTSA FARS police-coded number (~800 deaths/year), and the gap reflects a structural limitation: drowsiness leaves no roadside test and rarely appears in a crash report unless the driver self-reports falling asleep, which the most severely impaired drivers often cannot do.
The per-trip risk picture comes from Tefft’s 2018 case-control analysis in Sleep, drawn from the NHTSA National Motor Vehicle Crash Causation Survey. Compared with drivers who slept 7–9 hours, the adjusted odds of being culpable for a crash rose to 1.3× at 6–7 hours, 1.9× at 5–6 hours, 2.9× at 4–5 hours, and 15.1× at less than 4 hours (95% CI 4.2–54.4). For a driver who routinely operates the car after 4–5 hours of sleep — a not-uncommon pattern for shift workers, new parents, or anyone driving home from a long day — applying a conservative 4× per-trip fatal-crash multiplier to the sober baseline of roughly 1 in 50,000 trips yields about 1 in 12,500 per impaired trip. Compounding 480 such trips over 40 years of driving (monthly frequency) gives a cumulative lifetime probability of approximately 1 in 26 of involvement in a fatal crash specifically caused by their own fatigue.
The calibration failure is structural and similar in shape to the alcohol case. Dawson and Reid’s 1997 Nature experiment established that 17 hours of sustained wakefulness produces hand-eye coordination impairment equivalent to 0.05% BAC, and 24 hours equivalent to about 0.10% BAC — the same level at which the relative crash-risk multiplier is roughly 5–6×. The cultural acceptance of driving “a bit tired” has no analogue in the alcohol case, even though the impairment is on the same order. The naturalistic dashcam evidence is unambiguous: drivers in the seconds before a fatigue-related crash rarely exhibit the obvious warning signs the public-health literature emphasises; they typically appear engaged with the road right up to the moment of a microsleep or attentional lapse, which means the in-cabin sensation of “I’d know if I were too tired” is not a reliable detector.
Claim ledger
Every number below is what each source reported, with the verbatim quote we relied on and how we arrived at our figure. Click any link to verify directly.
-
[1] AAA Foundation for Traffic Safety — Drowsy Driving in Fatal Crashes, United States, 2017–2021
Drowsy Driving in Fatal Crashes, United States, 2017–2021- Statistic
An estimated 17.6% of all fatal crashes in the United States from 2017 through 2021 involved a drowsy driver — approximately 29,834 fatalities over the five-year period (~6,000 per year), roughly ten times the number identified by police-reported FARS coding alone.- Excerpt
“"An estimated 17.6% of all fatal crashes in the years 2017–2021 involved a drowsy driver." ”
- Source data from
- 2024-03-01
- Accessed
- 2026-05-25 · archived copy
- Calculation
- AAA Foundation for Traffic Safety 2024 report applied a model trained on NHTSA's Crash Investigation Sampling System (CISS), in which trained investigators reach detailed determinations of crash causation, to the FARS census of fatal crashes. The 17.6% figure (≈6,000 deaths/year over five years, ≈29,834 total) is the population-level anchor for the gap between police-reported and true drowsy-crash prevalence. Used here to establish that drowsy driving is roughly comparable in magnitude to alcohol-impaired driving (which kills ~12,000/year per NHTSA FARS).
-
[2] Tefft, B.C. — Sleep (Oxford Academic) — Acute sleep deprivation and culpable motor vehicle crash involvement
Acute sleep deprivation and culpable motor vehicle crash involvement- Statistic
Compared with drivers who slept 7–9 hours in the 24 hours before crashing, adjusted odds of culpable crash involvement were 1.3× for 6–7 hours, 1.9× for 5–6 hours, 2.9× for 4–5 hours, and 15.1× (95% CI 4.2–54.4) for less than 4 hours.- Excerpt
“"Drivers who reported having slept for 4 hours and less than 4 hours in the 24 hours before crashing had 2.9 (95% CI = 1.4 to 6.2) and 15.1 (95% CI = 4.2 to 54.4) times the odds, respectively, of having been culpable for their crashes, compared with drivers who reported 7–9 hours of sleep." ”
- Source data from
- 2018-09-01
- Accessed
- 2026-05-25 · archived copy
- Calculation
- Tefft 2018 is the peer-reviewed publication in Sleep of analysis drawn from the NHTSA National Motor Vehicle Crash Causation Survey (NMVCCS), a nationally representative sample of 5,470 crashes. The case-control design uses non-culpable drivers as the control arm, isolating the effect of sleep deprivation from time-of-day and exposure confounds. The 2.9× and 15.1× odds ratios for the <5-hour buckets are the multipliers used in the lifetime calculation: a conservative ~4× for "moderately drowsy" (4–5h sleep) anchors the headline ~1 in 26 estimate; the 15× high-end multiplier drives the upper uncertainty bound.
-
[3] National Highway Traffic Safety Administration (NHTSA) — Drowsy Driving: Avoid Falling Asleep Behind the Wheel
Drowsy Driving: Avoid Falling Asleep Behind the Wheel- Statistic
NHTSA estimates that in 2017, 91,000 police-reported crashes involved drowsy drivers, resulting in approximately 50,000 injuries and nearly 800 deaths — figures NHTSA explicitly describes as an underestimate.- Excerpt
“"NHTSA estimates that in 2017, 91,000 police-reported crashes involved drowsy drivers. These crashes led to an estimated 50,000 people injured and nearly 800 deaths." ”
- Source data from
- 2023-01-01
- Accessed
- 2026-05-25 · archived copy
- Calculation
- The NHTSA FARS-based figure (~800 deaths/year) is the police-reported lower bound. NHTSA itself, the AAA Foundation, and the National Sleep Foundation all agree this number undercounts the true total by at least 8–10×, primarily because police rarely record drowsiness as a crash factor unless the driver self-reports falling asleep. The gap between this number and the AAA 2024 estimate (~6,000 deaths/year) is the under-reporting factor for drowsy driving.
-
[4] Dawson, D. & Reid, K. — Nature 388:235 — Fatigue, alcohol and performance impairment
Fatigue, alcohol and performance impairment- Statistic
Hand-eye coordination performance after 17 hours of sustained wakefulness is equivalent to that observed at a blood alcohol concentration of approximately 0.05%; after ~24 hours of wakefulness, equivalent to roughly 0.10% BAC.- Excerpt
“"After 17 hours of sustained wakefulness cognitive psychomotor performance decreased to a level equivalent to the performance impairment observed at a blood alcohol concentration of 0.05%." ”
- Source data from
- 1997-07-17
- Accessed
- 2026-05-25 · archived copy
- Calculation
- Dawson & Reid 1997 is the canonical wakefulness-to-BAC equivalence reference (n=40, within-subjects counterbalanced design). Used here to anchor the intuition that "very tired" is not a lay descriptor but a quantifiable impairment state. The 17-hour and 24-hour thresholds are cited verbatim by NHTSA, CDC, and the National Sleep Foundation as the basis for their public-health framing of drowsy driving as comparable to alcohol-impaired driving.