One of the central and ongoing concerns of educators is assessing the effects and outcomes of particular teaching practices and learning conditions. Our interest lies not only in the extent to which students have learned something, but also in the extent to which what they have learned is related to what has been taught and how it has been taught. The processes that constitute that relationship are seldom directly accessible, however, either to those immediately and actively engaged as participants or to observers. The case will be developed here that the salient issue for both teachers and educational researchers is re-collecting the empirical grounds of pedagogical practices (i.e., interactional practices) which can provide for the possibility of reasoned reflection and review of those practices. Yet what we can know is irremediably shaped by how it is that we know. Despite the fact that this constraint frequently remains unaddressed in qualitative inquiry, the relationships between teaching and learning have been examined from various ethnographic and interactional perspectives, with more or less analytically interesting and useful results. However, the difficulties deriving from our epistemological stance remain. The central focus of the following pages is to raise these issues as difficulties faced by persons concerned with making assessments of cognitive skills, and to recommend an alternative conception of method, or, more accurately set of methods (i.e., Ethnomethodology) (EM) which can provide a perspective from which to address these difficulties and a means of reflecting on them in our practice. The latter sections of the paper provide a working example of method-in-practice and suggest the differences that EM could make to our assessments of one dimension of problem-solving skills reflection.