We show that the proliferation of armed groups increases the amount of organized political violence. The natural death of a tribal leader provides quasi-experimental variation in the number of armed groups across districts in Pakistan. Employing event study designs and IV-regressions allows us to isolate the effect of the number of armed groups on political violence from locational fundamentals of conflict, e.g., local financing and recruiting opportunities or government capacity. In line with the idea that armed groups compete for resources and supporters, we estimate semi-elasticities of an additional armed group on political violence ranging from 50 to 60%. Introducing a novel proxy for government counter-insurgency efforts enables us to show that this increase is driven by insurgency groups and not the state. Moreover, we show that groups splitting-up compensate for their capacity loss by switching to non-capital intensive attacks.