Why are some countries more prone to civil wars than others? A prominent explanation in the literature looks to an opportunity-based understanding of armed conflict based either on economic factors or on state power factors. Distinguishing between the effects of income versus effects of the state has been difficult to date because of problems of data availability. We introduce develop, operationalize, and validate a new measure of the state that we term state presence, and we deploy this measure in statistical analysis to tease apart the two explanations. We find robust evidence that state capacity has an independent effect on civil war onset, and only weak evidence for the economic explanation. This result implies that policy interventions for preventing armed conflict should focus on strengthening the state.