Country-level indicators such as gross domestic product, bureaucratic quality, and military spending are frequently used to approximate state capacity. These factors capture the aggregate level of state capacity, but do not adequately approximate the actual distribution of capacity within states. Intra-state variations in state capacity are critical to understanding the relationship between state capacity and civil war. We offer nighttime light emissions as a measure of state capacity to differentiate its impact on civil war onset within the country from its effect at the country level. We articulate pathways linking the distribution of nighttime light with the expansion of state capacity, and validate our indicator against other measures at different levels of disaggregation across multiple contexts. Contrary to conventional wisdom, we find that civil wars are more likely to erupt where the state exercises more control. We advance three mechanisms accounting for this counterintuitive finding: rebel gravitation, elite fragmentation, and expansion reaction. In the first scenario, state presence attracts insurgent activities. In the second, insurgents emerge as a result of the fragmentation of political elites. In the third, anti-state groups react violently to the state penetrating into a given territory. Finally, we validate these mechanisms using evidence from sub-Saharan Africa.