Existing research on how democratization may influence the risk of civil war tends to consider only changes in the overall level of democracy and rarely examines explicitly the postulated mechanisms relating democratization to incentives for violence. The authors argue that typically highlighted key mechanisms imply that elections should be especially likely to affect ethnic groups' inclination to resort to violence. Distinguishing between types of conflict and the order of competitive elections, the authors find that ethnic civil wars are more likely to erupt after competitive elections, especially after first and second elections following periods of no polling. When disaggregating to the level of individual ethnic groups and conflicts over territory or government, the authors find some support for the notion that ethno-nationalist mobilization and sore-loser effects provoke postelectoral violence. More specifically, although large groups in general are more likely to engage in governmental conflicts, they are especially likely to do so after noncompetitive elections. Competitive elections, however, strongly reduce the risk of conflict.