We study how armed conflicts affected educational outcomes in Rwanda during the nineties, relying on two waves of population census data and on a difference-in- differences identification strategy. Our results indicate that the conflicts caused on average a 22% drop in schooling attainments, corresponding to about one year less of education, and that the drop was relatively larger for girls. Primary and secondary schooling attainments were both affected, although through different channels. While increased drop- outs and school delays explain the drop in primary schooling, secondary schooling was mainly affected by a drop in enrollments. Finally, in a within-country analysis, we find no robust link between subnational variations in the drop in schooling and the intensity of any specific form of violence, despite the refined geographical measures at our disposal and a large set of checks. We present possible explanations for the observed patterns and provide related policy implications.