Increased attention paid to inter-group genetic variability following completion of the Human Genome Project has provoked debate about race as a category of classification in biomedicine and as a biological phenomenon at the level of the genome. Philosophers of science favor a metaphysical approach relying on natural kind theorizing, the underlying assumptions of which structure the questions asked. Limitations arise the more metaphysically invested and less attuned to scientific practice these questions are. Other questions—arguably, those that matter most socially and politically—remain unasked, not merely overlooked but systematically ignored and even foreclosed. Race fails as a postulated natural kind because it fails to meet expectations that as a category of classification it furnish an authoritative taxonomy that by depicting fundamental divisions in nature is conducive to fulfilling far-ranging explanatory aims. Racial, ethnic, and other group designations may nonetheless be projectible insofar as they support inductive inferences in biomedicine, but this does not render them any less social. Indeed, the statistical, contingent, accidental, localized, and interest-relative bases of such inferences serve to undercut the dichotomizing of race as either biological reality or social construct and favor the adoption of a pragmatic approach.