Few events give the opportunity to observe the full range of human behavior as wars do. In the case of civil wars in ethnically-mixed societies, the distribution of violence across various segments of the population can provide evidence on the extent and nature of discrimination. As in the case of markets, identifying discrimination in the warplace is challenging. There is uncertainty on the reconstruction of events as well as the rationale behind the violence. We use a unique data set collected by the Peruvian Truth and Reconciliation Commission on war crimes during the 1980’s to show that there is evidence of taste-based discrimination by agents of the state towards ethnic minorities and women. The evidence is robust to different assumptions on the logic of repression and missing data problems.