The higher-level phylogeny of eutherian mammals is a source of controversy because of conflict between phylogenetic analyses emphasizing morphology and those emphasizing DNA sequences as character sources. Each data partition supports clades that receive little support from the other partition and, in places, the two partitions are in direct conflict. Potential explanations for this discrepancy are problems in existing morphologic data sets, particularly their reliance on extant placental mammals to investigate a radiation that occurred tens of millions of years ago. This study presents a new morphological data set that focuses on dental and postcranial morphology. A large taxonomic sample of primarily Paleocene and Eocene eutherians is used, with the goal of avoiding some of the problems associated with homoplasy in extant taxa. Analysis of this matrix, with and without existing cranial data and a molecular constraint are presented. Results are most consistent with existing morphological analyses, with Templeton tests indicating that trees constrained to agree with a composite molecular topology present a significantly worse fit to the character data than do unconstrained trees. There is some rapprochement with molecular topologies, including recognition of a clade including Archonta and Glires similar to molecular Euarchontoglires and recognition of a monophyletic assemblage of ungulate-like African mammals. This study provides new data on the higher-level relationships of numerous extinct higher-level taxa. Among the more significant results are a challenge to the monophyly of Carnivoramorpha and recognition of a substantially monophyletic Leptictimorpha that may form the xenarthran stem taxon. A novel result unites several supposedly distantly related taxa (amphilemurid erinaceomorphs; plagiomenid archontans; apheliscid condylarths) with macroscelideans. Overall, this study does not unambiguously support alliance of most extinct lipotyphlans with living families. Support for a monophyletic clade including Pantodonta and Tillodontia, in concert with similar results in previous studies, favors recognition of a new superordinal clade for these groups. The results of this study are inconsistent with molecular hypotheses concerning the timing and biogeography of the placental radiation, instead supporting the morphological view that this radiation occurred on northern continents in association with the K/T boundary.