24Central to understanding early animal evolution are the questions of when and how many times in the ancestry of extant animals "eumetazoan" traits -nervous and digestive systems, striated musculature, and potentially defined mesoderm or its precursors -have arisen. The phylogenetic placement of the only two major animal clades lacking these traits, poriferans 28 (sponges) and placozoans, is crucial to this point, with the former having received much attention in recent years, and the latter relatively neglected. Here, adding new genome assemblies from three members of a previously unsampled placozoan lineage, and including a comprehensive dataset sampling the extant diversity of all other major metazoan clades and choanoflagellate outgroups, we 32 test the positions of placozoans and poriferans using hundreds of orthologous protein-coding sequences. Surprisingly, we find strong support under well-fitting substitution models for a relationship between Cnidaria and Placozoa, contradicting a clade of Bilateria + Cnidaria (= Planulozoa) seen in previous work. This result is stable to Dayhoff 6-state recoding, a strategy 36 commonly used to reduce artefacts from amino acid compositional heterogeneity among taxa, a problem to which the AT-rich Placozoa may be particularly susceptible. We also find that such recoding is sufficient to derive strong support for a first-splitting position of Porifera. In light of these results, it is necessary to reconsider the homology of eumetazoan traits not only between 40 ctenophores and bilaterians, but also between cnidarians and bilaterians. Whatever traits are homologous between these taxa must also have occurred in the evolutionary history of Placozoa (or occur cryptically in modern forms), and the common ancestor of Cnidaria and Bilateria may extend deeper into the Precambrian than is presently recognized.