Phylogenetic relationships are inferred principally from two classes of data: morphological and molecular. Most current phylogenies of extant taxa are inferred from molecules, and when morphological and molecular trees conflict the latter are often preferred. Although supported by simulations, the superiority of molecular trees has never been assessed empirically. Here we test phylogenetic accuracy using two independent data sources: biogeographical distributions and fossil first occurrences. For 48 pairs of morphological and molecular trees, we show that molecular trees are, on average, significantly more biogeographically congruent than their morphological counterparts. We also report an increase in the biogeographical congruence of phylogenies over research time. We find no significant differences in stratigraphical congruence between morphological and molecular trees. These findings have implications for understanding homoplasy in morphological data sets, the utility of morphology as a test of molecular hypotheses, and the difficulty of analysing fossil groups for which molecular data are unavailable.