The living and extinct flightless rails of the Pacific are among the most species-rich examples of parallel evolution in vertebrates. The “typical” rails of this region comprise a diverse assemblage of long-billed species variously placed in the genera Rallus, Lewinia, Nesoclopeus, Gallirallus, Habropteryx, Tricholimnas, Aramidopsis, Amaurornis, Eulabeornis, and Habroptila. I present a phylogenetic hypothesis for this group based on Bayesian and maximum likelihood analyses of 12S, control region, and cytochrome-b data obtained from museum specimens (frozen tissues, toe pads from study skins, and bones from archaeological sites) of living and extinct species. All previously recognized genera are either monotypic or non-monophyletic, and I advocate lumping nearly all species into a broadly defined Gallirallus sensu lato. Volant species are not paraphyletic with respect to nearly all flightless species. Instead flightless species branch off in rapid succession from lineages leading to extant volant species. The nesting of the flightless species G. pendiculentus with G. philippensis suggests that the flightless condition may evolve prior to reproductive isolation. A locally calibrated relaxed molecular clock indicates that species from Oceania evolved only within the last 400,000 years, supporting the hypothesis that speciation proceeds rapidly in flightless rails. These results help resolve a long-standing taxonomic quagmire and have important implications for Pacific biogeography and the tempo and mode of speciation in island birds.