While it has been known for some time that crashes can result from the driver falling asleep at the wheel, this issue has received less attention in traffic safety programs than the role of alcohol or speed of the vehicle. The present study was done to investigate the characteristics of crashes attributed to the driver being asleep. The study utilized the database at the Highway Safety Research Center at the University of North Carolina that is based on the uniform crash reporting system in that state. Over the years 1990-1992, inclusive, there were 4333 crashes in which the driver was judged to be asleep but not intoxicated. The crashes were primarily of the drive-off-the-road type (78% of the total) and took place at higher speeds (62% in excess of 50 mph). The fatality rate was of similar magnitude to that in alcohol-related crashes with fatalities in 1.4% of such crashes (alcohol crashes had fatalities in 2.1%). The crashes occurred primarily at two times of day--during the nighttime period of increased sleepiness (midnight to 7.00 a.m.) and during the mid-afternoon "siesta" time of increased sleepiness (3.00 p.m.). These crashes occurred predominately in young people. Fifty-five percent of these were in individuals 25 years of age or younger, with a peak age of occurrence at age 20 years. Sleepiness may play a role in crashes other than those attributed by the police to the driver being asleep. Determining the magnitude of this role is a challenge to the traffic safety community.