The molecularization of race thesis suggests geneticists are gaining greater authority to define human populations and differences, and they are doing so by increasingly defining them in terms of U.S. racial categories. Using a mixed methodology of a content analysis of articles published in Nature Genetics (in 1993, 2001, and 2009) and interviews, we explore geneticists' population labeling practices. Geneticists use eight classification systems that follow racial, geographic, and ethnic logics of definition. We find limited support for racialization of classification. Use of quasi-racial "continental" terms has grown over time, but more surprising is the persistent and indiscriminate blending of classification schemes at the field level, the article level, and within-population labels. This blending has led the practical definition of "population" to become more ambiguous rather than standardized over time. Classificatory ambiguity serves several functions: it helps geneticists negotiate collaborations among researchers with competing demands, resist bureaucratic oversight, and build accountability with study populations. Far from being dysfunctional, we show the ambiguity of population definition is linked to geneticists' efforts to build scientific authority. Our findings revise the long-standing theoretical link between scientific authority and standardization and social order. We find that scientific ambiguity can function to produce scientific authority.