Cephalodiscus gracilis Harmer, 1905 is a semi-sessile deuterostome that shares with fish-like chordates pharyngeal gill slits and a dorsally situated brain. In order to reveal structures potentially homologous among deuterostomes and to infer their functional roles, we investigated the axial complex, associated blood vessels and structures of C. gracilis using transmission electron microscopy, light microscopy, and digital 3D reconstructions. We describe the smooth, bipartite cephalic shield retractor muscles that originate as solid compact muscles and fan out to traverse the protocoel as individual muscle cells. The axial complex consists of a cap-shaped coelomic sac, the pericardium that surrounds the central heart. The pericardium is constituted of myoepithelial cells, with the cells facing the heart being thicker and richer in myofilaments. A prominent dorsal median blood vessel opens into the heart, which gives rise to a short median ventral vessel that opens into the paired glomeruli connected to the ventral side of the stomochord. The tip of the curved stomochord rests precisely above the connection of the dorsal median vessel with the heart, a position that would allow the stomochord to function as a valve facilitating unidirectional blood flow. Glomeruli are lined by podocytes of the spacious protocoel and are considered to be the site of ultrafiltration. Two pairs of blood vessels enter the median dorsal blood vessel from the tentacles. The median dorsal blood vessel is separated from the brain by a thin basement membrane. This arrangement is consistent with the hypothesis that blood vessels in the tentacles increase oxygen supply for the brain. Based on detailed similarities, the heart–glomerulus complex of C. gracilis is considered homologous with the heart–glomerulus complex in Rhabdopleura spp., and Enteropneusta, and the axial complex in Echinodermata. In addition, we hypothesize homology to the excretory complex including Hatschek’s nephridium in Cephalochordata. Thus, the heart–glomerulus complex does not support a sister-group relationship between Echinodermata and Hemichordata, whereas the organization of the cephalic shield retractor muscles is consistent with the evolution of pterobranchs within enteropneusts.