Most nations have experienced an internal armed conflict since 1960. The past decade has witnessed an explosion of research into the causes and consequences of civil wars, belatedly bringing the topic into the economics mainstream. This article critically reviews this interdisciplinary literature and charts productive paths forward. Formal theory has focused on a central puzzle: why do civil wars occur at all when, given the high costs of war, groups have every incentive to reach an agreement that avoids fighting? Explanations have focused on information asymmetries and the inability to sign binding contracts in the absence of the rule of law. Economic theory has made less progress, however, on the thornier (but equally important) problems of why armed groups form and cohere, and why individuals decide to fight. Likewise, the actual behavior of armed organizations and their leaders is poorly understood. On the empirical side, a vast cross-country econometric literature has aimed to identify the causes of civil war. While most work is plagued by econometric identification problems, low per capita incomes, slow economic growth and geographic conditions favoring insurgency are the factors most robustly linked to civil war. We argue that microleve analysis and data are needed to truly decipher war's causes, and understand the recruitment, organization, and conduct of armed groups. Recent advances in this area are highlighted. Finally, turning to the economic legacies of war, we frame the literature in terms of neoclassical economic growth theory. Emerging stylized facts include the ability of some economies to experience rapid macroeconomic recoveries, while certain human capital impacts appear more persistent. Yet econometric identification has not been adequately addressed, and there is little consensus on the most effective policies to avert conflicts or promote postwar recovery. The evidence is weakest where it is arguably most important: in understanding civil wars' effects on institutions, technology, and social norms.