Although popularly perceived as a positive force important for objectives such as economic development and democracy, social capital may also be linked to less desirable outcomes. This article highlights a dark side to social capital by pointing to its role in a particularly pernicious phenomenon: genocidal violence. Drawing on a survey of residents from one community that experienced violence during Rwanda's 1994 genocide, I show that individual participation in the violence was partly determined by the features of residents' social networks. Perpetrators possessed larger networks in general and more connections to other perpetrators in particular. The quality as well as quantity of connections also mattered. Strong ties generally, and kinship and neighborly ties specifically, were strong predictors of participation. In contrast, possession of countervailing ties to nonparticipants did not reduce a resident's likelihood of participation. Drawing on in-depth interviews to explore the possible mechanisms behind these findings, I suggest participants' networks fulfilled functions of information diffusion, social influence, and behavioral regulation. More broadly, the findings suggest the importance of social structure and social interaction for participation in collective violence. Relational data should complement individual attribute data in predicting participation. The findings also suggest, contrary to the neo-Malthusian interpretation, that the role played by Rwanda's extraordinarily high population density in the violence may have been more sociological than ecological in origin. The diffusion, influence, and regulatory effects of social connections are likely to be amplified in communities where individuals live in close spatial proximity to each other.