The structure of the proboscis in different polychaetes is described and compared with the structures derived from the stomodeum in those polychaetes in which no proboscis is to be found. Evidence is presented to support the contention that the structures which arise from the stomodeum are of phylogenetic value and that as a result suggestions for a less arbitrary classification of the Polychaeta than those commonly used are made. It is suggested that those worms with a ventral proboscis represent mainly separate lines of evolution from a primitive stock provided with a ventral buccal organ similar to that found today in archiaimelids, and that this buccal organ has been retained with relatively little modification in some deposit feeders such as ampharetids and terebellids. It is also suggested that an axial proboscis was evolved at an early stage of evolution and has diverged into two main types, one without and one with a gizzard-like pharynx; the first, in sand and mud eaters in which the simpler proboscis tends to be lost on transition from deposit to suspension feeding (as in chaetopterids and sabellariids), the second, in worms of more diverse but predominantly predatory or scavenging habit.