After the Nazi Holocaust, the international community vowed to prevent genocide from occurring in the future. Yet, genocide has continued to occur. Accordingly, this study seeks to better understand why and how genocide takes place. I ask two key questions: 1) What are the causes of genocide at societal, state, and international levels? and 2) What accounts for temporal and regional variation in violence within genocides? To assess what leads to genocide, I conduct an event history analysis of the preconditions of genocide in all countries over the last 50 years. This quantitative analysis examines factors associated with the onset of genocide at the societal level (such as ethnolinguistic diversity), state level (such as type of government), and the international level (such as trade), finding that factors at each level must be considered in order to understand why genocides take place and that civil wars are the strongest predictors of genocide. While the event history analysis treats genocide as a single event, viewing genocide as an undifferentiated event misses opportunities to better understand the violence. Thus, the second part of this dissertation draws upon three case studies to analyze regional and temporal variation in genocidal violence in Rwanda, Bosnia-Herzegovina, and Sudan. I rely upon quantitative models to test how numerous factors drawn from genocide studies, the study of political and ethnic violence, and criminology--such as ethnic diversity, resource scarcity, unemployment levels, education levels, or the presence of certain armies--influence the onset and magnitude of certain forms of violence at meso levels. I also conducted fieldwork and 113 interviews with survivors, scholars, and other witnesses. Overall, I find that the factors associated with regional and temporal differences in violence vary based on who the perpetrators are and how they are organized. In Rwanda, members of the community who were not part of previously organized formal groups participated in the violence. As such, criminology's social disorganization theory--which argues that community cohesion influences crime rates--helps explain variation in this violence. In Bosnia-Herzegovina and Darfur, however, previously organized armies and militias generally committed the violence. Accordingly, strategic concerns dictated patterns.