Study of vocalizations of Herpsilochmus antwrens in northeastern Brazil revealed that the traditional H. pileatus complex is comprised of not two but three species-level taxa. We show that the systematics and biogeography of the group have been based on an unnamed species that has been called H. pileatus for most of this century. Here, we name this “new” species and conclude that its several morphological and vocal distinctions clearly set it apart from other members of the complex. True H. pileatus, as determined through comparison of the lectotype with recently collected specimens and from the distribution of recordings, is confined to southern coastal Bahia. Morphological and vocal analyses suggest that H. pileatus and H. atricapillus are each other's closest relatives. We provide a simple key to diagnostic plumage and mensural characters to permit identification of museum specimens. An extensive series of specimens demonstrated that H. atricapillus includes two male morphs. Geographic overlap of pure morphs in southern Goiás, and the existence of some apparent intermediates in this general area, suggest secondary contact of two weakly differentiated populations, perhaps along complex habitat gradients. Analysis of recordings of H. atricapillus disaggregated geographically and by male morph revealed no diagnostic (by criteria of Isler et al. [1998]) variation within that species, although average measures of overall pace and pace of three sections of loudsongs differed between the morphs by approximately 20%, which we interpret as another indication of incipient differentiation. Revision of the Herpsilochmus pileatus complex provides a case in point for the fundamental role in conservation of both field and taxonomic research and demonstrates the importance of examination of type specimens and inclusion of topotypical samples (whether anatomical, biochemical, or tape recordings) in studies of taxonomy and systematics.