This paper examines how exposure to neighborhood homicide affects school performance, fear of crime, and self-reported offending among participants in UMSL's Comprehensive School Safety Initiative, a longitudinal study of middle school students in the St. Louis region. We operationalize neighborhood homicide as any homicide occurring within a one-mile radius of students' home addresses. We exploit natural variation in the timing of survey administration to address selection-artifacts associated with residential sorting. Specifically, we compare the survey responses of students who experienced a homicide in the immediate vicinity of their home in the one-month period leading up the administration of the survey to students who did not experience a homicide in that period but would in the month following survey administration. This approach allows us to hold constant endogenous factors associated with the likelihood that respondents live near areas with an elevated probability of experiencing homicide as well as their subsequent scores on our outcomes of interest.