Undergraduates (N = 330) estimated their risk for aversive events (e.g., motor vehicle accident), in which we varied personal control (e.g., driver = self or other) and bogus base rates (high vs. low). Base rate data influenced risk estimates, but people underestimated their risk if they were the agent (e.g., the driver) rather than if another person was the agent. Justifications for estimates suggested that the base rate was perceived to be the risk for the average person. Most people then adjusted their estimates for a negative outcome downward because they believed that they were better than average on a skill (e.g., driving skill) that was perceived to be causally related to the event.