Power sharing is often purported to lead to civil peace, though its effects are disputed. We identify three types of power sharing—inclusive, dispersive, and constraining—and analyze their mechanisms of power allocation. We argue that constraining arrangements, which limit the power of a party or social group, are most likely to protect vulnerable groups. Constraining power-sharing institutions, such as guarantees of civil liberties and independent judiciaries to protect them, create checks against governmental repression of minority groups and ordinary citizens, whereas other types of power sharing focus on political elites. Constraining power-sharing institutions limit incentives for mass mobilization and raise the costs of conflict, thus reducing the probability of civil war. Drawing on a new global data set, we examine the effects of power sharing on the onset and recurrence of civil war. In contrast to prior findings, our results show that only constraining power-sharing institutions have a pacific effect.