We challenge assumptions that have guided much research and policy aimed at understanding and reducing medical practice variation. Paramount is the focus on doctors as the cause of variation to the neglect of other possible influences. Some research literature suggests that patients, families, and the community context of practice may also influence treatment decisions. Failure to question present assumptions, despite weak evidence in support of them, may account for inability to explain persistent practice variation, develop appropriate implementable guidelines, or anticipate the effect on treatment decisions of greater patient involvement. In this paper, we discuss the weak response to the NIH Consensus Conference on early stage breast cancer because it may have reflected these problems. We urge a more complex and more empirical approach in explaining treatment choice and guidelines sensitive to the potential for value differences.