Recent theory in anthropology has increasingly been concerned with issues of power. Anthropology also has a long history of interest in variation in cultural knowledge, which, we argue, benefits from attention to power relations. To show this, we examine perceptions of breast cancer risk factors among physicians. Although physicians share a general cultural model of breast cancer risk factors, variation exists, especially between university-based physicians and community-based physicians. The nature of the work performed in these two settings influences the acquisition of various sources of information and frames what is considered valid information. Similar to Foucault's argument, we find that physicians working in a university setting are more disciplined in discussing their perceptions of breast cancer risk factors, compared to community-based physicians, who move away from the centers of knowledge and power (universities).Key Words: power relations; variation in medical knowledge; knowledge, power and medicine; consensus analysis, breast cancer risk factorsIn a critique of the representation of biomedicine as "monolithic and uniform," DiGiacomo (1992) states that "this sort of broad-stroke generalization seems incautious, given the cultural diversity to be found even in the 'advanced industrial world'." Rather than assuming biomedicine is monolithic, anthropologists studying biomedicine as a cultural system have explored how biomedicine is socially, culturally, and historically constructed (Good and Good 1993; Lindenbaum and Lock 1993; Amarasingham Rhodes 1991;Good, et al. 1990; Lock and Gordon 1988; DiGiacomo 1987). In addition, researchers have examined how physicians have maintained their dominant positions, vis-a-vis non-physicians and peripheral medical practitioners, through the use of "technology, engaging in scientific and objective research, and the ability to measure biological aspects of illness" (Lock