Cross-national empirical studies of corruption commonly find that nations in which women play a greater role in economic and public life suffer less corruption. This finding has been controversial in that measures of women's participation in the labour force and politics are potentially endogenous. This study uses an aspect of national ancestral geography as an instrumental variable towards estimating the true causal effect of gender upon corruption. The ensuing estimates indicate that ordinary least squares estimates of the coefficients of regressors measuring women's economic and political influ-ence, in regressions in which measured corruption is the dependent variable, are substantially biased.