Foucault's work has given rise to increased methodological sensitivity of the political dangers associated with traditional qualitative approaches in the social sciences. There is a growing awareness that the widespread use of the research interview is not indicative of a deepening insight into the workings of culture, but is part of a broader social technology for its reproduction. In an effort to re-imagine interview methodology, scholars have read Foucault to suggest the need for greater attention to the active co-construction of research conclusions arising from interview based research. This has led in turn to the view that post modern approaches produce localized, temporally specific knowledge that fails to shed light on deeper, more enduring social structures. This paper questions these interpretations of Foucault's work, arguing that they fail to accurately represent his genealogical method or to consider its implications for research ethics. Foucault rejects a view of knowledge as emerging from the active social constructions of agents or of institutionalised ‘interests’. Rather, Foucault sees knowledge as an outcome, often accidental, of interrelated historical practices and discourses that produce the subjects and objects of social science discourse itself. The implications of Foucault's work for thinking about research ethics is not a return to authenticity or to analyses of social structure, but a rejection of the centralised, regulatory claims of an organised scientific discourse. The paper comprises a review of social science responses to post structural insights, coverage of the critical epistemological differences between Foucault's method and other key social theory paradigms, and a discussion of the critical ethical issues these differences raise for the social sciences.