This book examines how and why episodes of state violence begin and end. The central argument is that researchers should give more attention to the key policy question of how large-scale violent repression can be ended. The authors analyze repressive spells, or periods of time where a state deliberately engages in large-scale repression. They focus on decision makers who begin a campaign of state violence and argue that violent campaigns can become sustained through a kind of inertia. This leads them to expect that the determinants of onset may not be the same as those of termination. The bulk of the book consists of statistical analyses that examine the impact of various conditions on spell onset, escalation, termination, and recurrence. Throughout, the authors connect their findings to policy debates about the relative effectiveness of domestic efforts to end repression (through democratization and civil resistance) and international efforts (fostering civil resistance/democracy from abroad, economic sanctions, military interventions, and greater inclusion in international organizations). They supplement extensive statistical analysis with historical narratives of individual countries. Their results suggest that domestic conditions such as democracy, democratization, and armed conflict are the most consequential for beginning and ending spells, but they also reveal some interesting differences between the influences of various conditions on different phases of repressive spells. For example, membership in international organizations is unrelated to onset but increases the probability of termination. Perhaps most relevant for the policy community, coercive measures such as military interventions and economic sanctions, if they have any impact, make state violence more likely to begin and/or escalate, and less likely to end.