A brief excursion through the history of social medicine suggests that, at least in principle, epidemiology and anthropology are natural allies in the study of disease in human populations. In practice, however, this alliance has been limited and remains problematic. This article examines the possibilities for interdisciplinary research, taking cancer epidemiology as a case in point. I argue, on the basis of participant-observation over a period of nearly two years in the epidemiology department of a medical research institute in Catalonia (Spain), that bioscientific uses of the concept of culture have led, disappointingly, to its reification as "beliefs" and its incorporation into the naturalist epistemology of Western institutional medicine. The unfortunate consequence is the medicalization of culture understood as "difference," which often stands in for social class.