Many African civil wars are fought along ethnic lines, and new rulers often find themselves facing an armed insurrection along ethnic lines led by former comrades-in-arms. Why do rulers exclude certain ethnic groups if exclusion increases the risk of civil war? In this masterful book, Roessler solves this puzzle by putting forward a strategic, competing-risks model of ruler behavior. Roessler argues rulers face a coup d'etat-civil war trap: ethnic inclusion in the government may reduce the risk of civil war, but it does so by watering down the benefits to the members of the ruling coalition, increasing the risk of a coup d'etat by their co-conspirators. These calculations entail commitment problems, as members of the ruling coalition develop personalist networks within the security apparatus because none of the members can credibly commit not to usurp power via a coup. Building these networks – placing confidantes in positions of power – further generates security dilemmas, as coup-proofing behavior threatens other members of the coalition, incentivizing 'first strike' behavior. Thus, African leaders are likely to both purge their former comrades and face subsequent rebellion, preferring to fight in the bush rather than risk a palace coup. Roessler's approach is rooted in both selectorate theory and case studies of Sudan's Islamic Movement and the outbreak of war in Darfur. He tests the theory using large-n analysis – leveraging the Ethnic Power Relations data – and a case study of the breakdown of Laurent Kabila's ruling coalition and the resulting Second Congolese Civil War (1998–2003). Ambitious in scope and theoretically and historically nuanced, this book is among the most exciting contributions to the study of ethnic politics and civil war in a decade.