Many researchers are currently studying the distribution of genetic variations among diverse groups, with particular interest in explaining racial/ethnic health disparities. However, the use of racial/ethnic categories as variables in biological research is controversial. Just how racial/ethnic categories are conceptualized, operationalized, and interpreted is a key consideration in determining the legitimacy of their use, but has received little attention. We conducted semi-structured, open-ended interviews with 30 human genetics scientists from the US and Canada who use racial/ethnic variables in their research. They discussed the types of classifications they use, the criteria upon which they are based, and their methods for classifying individual samples and subjects. We found definitions of racial/ethnic variables were often lacking or unclear, the specific categories they used were inconsistent and context specific, and classification practices were often implicit and unexamined. We conclude that such conceptual and practical problems are inherent to routinely used racial/ethnic categories themselves, and that they lack sufficient rigor to be used as key variables in biological research. It is our position that it is unacceptable to persist in the constructing of scientific arguments based on these highly ambiguous variables.