A common finding in the empirical civil war literature is that population size and per capita income are highly significant predictors of civil war incidence and onset. This paper shows that the common finding of population size and per capita income having a significant average effect on civil war risk in a world sample breaks down once country- and year-specific unobservables are accounted for. However, for Sub-Saharan Africa there continues to be a highly significant average effect of population size and per capita income on civil war risk that is robust to the use of country- and year-fixed effects and instrumental variable techniques.