I reconsider the primacy of institutions over geography as an explanatory factor of cross-country differences in economic performance, which has recently been postulated by Acemoglu et al. (2001) and others. My estimates show that the reported missing direct performance effects of a measure of geography such as malaria prevalence are not robust to alternative specifications and samples. Unbiased estimates of the relative performance effects of institutions and malaria prevalence are difficult to obtain due to a lack of independent instrumental variables. Conditional on a restricted effect of institutions, my estimates suggest that malaria prevalence exhibits a large negative direct impact on economic performance, as postulated by Sachs (2003) and others.