Alcohol is associated both with motor vehicle crashes and with crime. We examine the role of alcohol in hit-and-run collisions, based on pedestrian fatalities recorded in the Fatal Accident Reporting System. In 19% of all pedestrian fatalities in 1989 and 1990, the driver left the scene. Time of the accident is available for all pedestrian fatalities. Strong evidence for the influence of alcohol on hit-and-runs is the greater proportion of hit-and-runs at night and during the weekend, two periods when drivers are more likely to be drinking. Half of hit-and-run motorists are eventually identified. Compared to those who remain at the scene, the identified hit-and-run motorists are far more likely to have had a previous arrest for driving while intoxicated. They are also disproportionately young and male, two factors associated with drinking and driving. Only a fraction of drivers are tested for blood alcohol concentration (BAC); those who left the scene are more likely to have a positive BAC. Two theories--the rational decision theory and the personality theory--help explain why drunk drivers are more prone to run after hitting a pedestrian.