Stereotypes lead people to over-react to information that confirms their priors and under-react to information that goes against their priors. I test a model of stereotyping using data on police officer assaults. I match data from the New York City Stop, Question, and Frisk program to data on police deaths to test whether police officers have different responses to shootings depending on the shooter's race. I find that when an officer is shot by a white civilian, there is no change in frisk, arrest, or use of force patterns. However, when an officer is shot by a black civilian, frisking and use of force against black civilians increases dramatically. I test alternative explanations such as rational updating, retaliation, and demographic segregation.