What explains why certain individuals participate in episodes of collective violence and others not? Differential selection into riots, communal violence, and ethnic massacres has often been explained in terms of individual attributes: age, gender, occupation, education, income. Using social network analysis, I present a relational theory of participation to complement the attribute-based approach. Ifindparticipationisafunctionofthecharacteristicsof(i)anindividual‟snetwork;(ii)the connections within this network; and (iii) the individual actor. Drawing on Rwanda‟s genocide, Icompare participants and non-participants in the violence from one community. I find first the size ofan individual‟s network mattered. Participants were better connected generally and to otherparticipants specifically. Second, the type and strength of connections also mattered. Kinship connections and stronger connections to other participants better predicted participation. Third, thereexisted a small core of “organizers” whose influence was due to their individual characteristics ratherthan their network characteristics.