Adequate sleep is critical for general healthy functioning. This paper studies how sleep affects fatal vehicle crashes. The sleep data is from American Time Use Survey (ATUS) and the vehicle fatality data is from Fatality Analysis Reporting System (FARS). Using sunset time as instrument, I find that a one-hour increase in monthly sleep leads to a decrease of about 0.035 fatal crashes per 100 million vehicle miles traveled (VMT), equivalent to a 2.4% reduction in fatalities in the short run. Alternatively scaled, an extra hour of daily sleep reduces one fatal crashes in 2,724 years if one drives 100 miles per day. However, I find there is no significant long run effect of sleep on fatal vehicle crashes.