Why are communal conflicts more deadly in some civil war-affected areas than in others? Despite most violent communal conflicts taking place in civil war-affected countries, knowledge on how armed groups exacerbate or restrain communal violence is limited. Combining insights from the communal conflict and rebel governance literature, I argue that communal conflict severity should be lower in areas controlled by armed groups that appeal to a broad domestic constituency than in areas controlled by armed groups that appeal to a narrow constituency. This is because inclusionary armed groups have greater incentives to develop institutions and practices that restrain communal violence than exclusionary groups. The study examines the argument through a process-driven natural experiment in western Côte d’Ivoire, where an externally imposed ceasefire line randomly subjected highly comparable communities to inclusionary or exclusionary armed group control. Drawing on extensive fieldwork, the study demonstrates that communal conflicts were more deadly in areas controlled by exclusionary militias than in areas controlled by an inclusionary rebel group, and offers in-depth evidence of how inclusionary rebels contributed to resolve communal conflicts short of violence. The findings contribute new insights on the drivers of violent communal conflict in civil war and on how armed actors mediate communal conflicts to win legitimacy.