Against the background of sex-specific and sex-comparative approaches in health research, this article aims to clarify to what extent the category sex/gender with its biological dimensions (sex) and social dimensions (gender) has systematically and conceptually been consider ed in epidemiology and which methods have been developed. Epidemiologic research has been criticized for routinely controlling statistically for age and sex but often ignoring aspects of gender. Inadequate consideration of sex/gender may result in systematic errors (gender bias), on the one hand, if sex/gender is ignored as an important variable, and, on the other hand, if differences between men and women are assumed when there are actually similarities. There are examples of adequate consideration of gender in exposure assessment, analysis of social position or modelling of interactions in current articles of scientific journals. How ever, epidemiologic reference books and textbooks as well as university training in epidemiology show that the category sex/gender has not been integrated with both dimensions sex and gender into the currently predominating thought style of epidemiology. For the further development of valid epidemiologic research clarification of terms, generation of unambiguous concepts and sophisticated statistical tools are necessary. This is the only way to succeed in analysing the complex interactions between sex-linked biology and gender relations.