Taxonomic diversity and morphological disparity are different measures of biodiversity that together can describe large-scale evolutionary patterns. Diversity measures are often corrected by extending lineages back in time or adding additional taxa necessitated by a phylogeny, but disparity analyses focus on observed taxa only. This is problematic because some morphologies required by phylogeny are not included, some of which may help fill poorly sampled time bins. Moreover the taxic nature of disparity analyses makes it difficult to compare disparity measures with phylogenetically corrected diversity or morphological evolutionary rate curves. We present a general method for using phylogeny to correct measures of disparity, by including reconstructed ancestors in the disparity analysis. We apply this method to discrete character data sets focusing on Triassic archosaurs, Cenozoic carnivoramorph mammals, and Cretaceous–Cenozoic euarchontogliran mammals. Phylogenetic corrections do not simply mirror the taxic disparity patterns, but affect the three analyses in heterogeneous ways. Adding reconstructed ancestors can inflate morphospace, and the amount and direction of expansion differs depending on the taxonomic group in question. In some cases phylogenetic corrections give a temporal disparity curve indistinguishable from the taxic trend, but in other cases disparity is elevated in earlier time intervals relative to later bins, due to the extension of unsampled morphologies further back in time. The phylogenetic disparity curve for archosaurs differs little from the taxic curve, supporting a previously documented pattern of decoupled disparity and rates of morphological change in dinosaurs and their early contemporaries. Although phylogenetic corrections should not be used blindly, they are helpful when studying clades with major unsampled gaps in their fossil records.