Avian genera characterized by sibling species with distinctive habitat preferences present an evolutionary enigma in view of the more commonplace occurrence of syntopic congeners that differ strikingly in colour and pattern. No existing theory has explained the evolutionary background that led to these differences. Here we propose that great phenotypic similarity among some groups of sibling species limits their coexistence and that clues to their radiation can be seen in patterns of geographical occurrence. To illustrate our thesis we focused on the New World flycatcher genus Empidonax, a group of 15 species notorious for their great phenotypic similarity. Using 3069 base pairs of mitochondrial DNA from four genes, we produced a complete molecular phylogeny that identified four clades, three of which represent close relatives. The fourth clade includes only E. virescens, which apparently has no close living relatives. The majority of species, including many distant relatives, are completely (58.1%) or essentially (6.7%) allopatric in breeding distribution and exhibit striking ecological segregation into distinctive climate–vegetation zones. Even where ranges overlap, occupancy of the same habitat by different species is rare. Phylogenetic and distributional patterns in Empidonax suggest a peripatric model of stepwise colonization and then range expansion of small groups of pioneers during glacial periods into initially enlarging, distinctive habitats destined to be widespread during interglacials. Vicariance is not indicated in the absence of barriers of appropriate age and geographical position. Rapoport's rule that northern species have larger ranges than southern species is strongly supported.