We study the effects of economic shocks on civil conflict at the subnational level using a panel dataset of 5689 administrative regions from 53 African countries with yearly observations from 1992 to 2010. We find that economic shocks, measured by nighttime light intensity and instrumented by lagged rainfall levels and droughts, increase the probability of civil conflict.