Phyla are defined by two sets of criteria, one morphological and the other historical. Molecular evidence permits the grouping of animals into clades and suggests that some groups widely recognized as phyla are paraphyletic, while some may be polyphyletic; the phyletic status of crown phyla is tabulated. Four recent evolutionary scenarios for the origins of metazoan phyla and of supraphyletic clades are assessed in the light of a molecular phylogeny: the trochaea hypothesis of Nielsen; the clonal hypothesis of Dewel; the set-aside cell hypothesis of Davidson et al.; and a benthic hypothesis suggested by the fossil record. It is concluded that a benthic radiation of animals could have supplied the ancestral lineages of all but a few phyla, is consistent with molecular evidence, accords well with fossil evidence, and accounts for some of the difficulties in phylogenetic analyses of phyla based on morphological criteria.