Thylacocephala are among the most problematic of arthropod fossils. Various authors have allied them with all manner of crustacean groups, including branchiopods, cirripedes, remipedes, and malacostracans. They have a very apomorphic body plan often marked by hypertrophy of the compound eyes, three pairs of large raptorial subchelate limbs, eight sets of well-developed phyllobranch gills, and from 8 to at least 16 posterior trunk somites bearing paddle-like limbs. They have been thought of as a distinct class composed of two orders, Concavicarida and Conchyliocarida, but membership within the orders varies according to authors, and no familial divisions have been proposed within the orders until now. This lack of taxonomic structure inhibits organization of available information concerning the paleoecology, paleogeography, and phylogenetic relationships of thylacocephalans. A working hypothesis for the higher taxonomy within the class is proposed here. This entails a redefinition of the two orders, and recognition of seven families, five of them new: Austriocarididae Glaessner, , Clausocarididae Arduini, (new status), Concavicarididae n. fam., Dollocarididae, n. fam., Microcarididae n. fam., Ostenocarididae n. fam. and Protozoeidae n. fam.