Sewall Wright's F ST is a mathematical test used to characterize differences between biological populations and identify causes of those differences. I discuss Cockerham and Weir's popular, empirically successful approach to statistical estimation of F ST, arguing that their assumption that actual populations are sampled from an infinite set of counterfactual populations (with a common ancestral population) supports a view of natural selection and genetic drift as distinct causes. I also argue that the way in which F-statistics and other statistical tests are applied to computer simulations in empirical research shows that selection and drift correspond to differences in objective, causal probabilities. © 2012 by the Philosophy of Science Association. All rights reserved.