This paper considers the application of Foucauldian perspectives within sociology. While Foucault's epistemology has generated novel historical and philosophical interpretations, when transposed to sociology, problems arise. The first of these concerns the association of knowledge and power, and the concept of 'discourse'. Foucault suggested that there are 'rules' of discursive formation which are extraneous to the 'non-discursive' realm of 'reality'. This formulation is consequently both deterministic and incapable of supplying explanations of why some practices become discursive which others do not. This determinism is reflected in some sociological analyses of embodiment, offering a model of the 'body' which is passive, and incapable of resisting power/knowledge. Secondly, Foucault's notion of the 'self moves to the other extreme, inadequately addressing the constraints which affect the fabrication of subjectivity. Sociological accounts do not always recognize the ambiguities which consequently result from efforts to use Foucauldian positions. It is argued that post-structuralists other than Foucault may offer more to sociology.