Quantitative genetic theory predicts that when populations diverge by drift the interspecific divergence (D matrix), calculated from species means, will be proportional to the average value of the additive genetic variance-covariance matrix, or G matrix. Most empirical studies in which this hypothesis has been investigated have ignored phylogenetic nonindependence among included taxa. Baker and Wilkinson (2003; also Revell et al. 2007) used a test for constraint in which the D matrix is calculated from phylogenetically independent contrasts (Felsenstein 1985) instead of directly from the species means. I use computer simulations to show that, on average, when the process of evolution is genetic drift, the divergence matrix calculated from independent contrasts (D IC) is more highly correlated with G than is the divergence matrix calculated ignoring phylogenetic nonindependence (D). This effect is more pronounced when speciation is initially slow but increases over time than when speciation decreases over time. Finally, when evolution is primarily by drift but phenotype space is bounded (as if by functional constraint) the average correlation is decreased between both G and D or DIC, however the correlation between G and DIC is much larger than between G and D. Although limited in scope, to my knowledge this is the first study to use individual-based quantitative genetic simulations in a phylogenetic context. © 2007 The Author(s).