The evolution of reproductive isolation within Coreopsis is investigated by integrating phylogenetic data with estimates of pollen viability of plants from inter- and intraspecific crosses. Three different models that predict F₁ fitness are compared. The first uses ITS pairwise distances, the second is based on phylogenetic branch lengths derived from DNA sequences, and the third elaborates on the second model by dividing branch length according to reconstructions of the evolution of life history. This is the first study to use phylogenetic branch-length estimates for predicting levels of reproductive isolation. Estimated branch lengths (model 2) predict hybrid fitness more accurately than simply genetic distance (model 1) but only very slightly. This is probably because the two variables are strongly correlated in Coreopsis. Prediction is substantially improved by allowing evolutionary rates to differ between annual and perennial branches (model 3). A bootstrapping procedure indicates that the life-history effect is statistically significant. The more rapid evolution of reproductive isolation within annual species of Coreopsis may be due to differing mechanisms of reproductive isolation, that is, chromosomal rearrangements rather than genetic incompatibilities.