This paper uses variation in victimization probabilities between individuals living in the same community to shed new light on the costs of crime. I use panel data from the Mexican Family Life Survey for 2002 and 2005 and look at the impact of withincommunity differences in victimization risk on changes in self-rated health as well as in measures of physical and mental health. My results from fixed effects and instrumental variable estimations point towards substantial negative health effects of actual victimization, which might help to the existence of compensating differentials in wages or house prices found in earlier studies.