Numerous authors have critiqued the use of race as an etiologic quantity in medical research. Despite this criticism, the use of variables encoding racial/ethnic categorization has increased in epidemiology, and most researchers agree that important variation in disease risk is captured by this classification system. Previous discussions have generally neglected to articulate guidelines for appropriate use of racial/ethnic information in etiologic research. The authors summarize the logical, conceptual, and practical problems associated with the "ethnic paradigm" as currently applied in biomedical sciences and offer a set of methodological recommendations toward more valid use of racial/ethnic classification in etiologic studies. These suggested guidelines address issues of variable definition, study design, and covariate control, providing a consistent foundation for etiologic research programs that neither ignore racial/ethnic disease disparities nor obfuscate the nature of these disparities through inappropriate analytical approaches. This methodological analysis of racial/ethnic classification as an epidemiologic quantity provides a formal basis for a focus on racism (i.e., social relations) rather than race (i.e., innate biologic predisposition) in the interpretation of racial/ethnic "effects."