SummaryWe have presented a review of some studies of interaction in nursing. The problem of analyzing interaction can be seen as a problem in the application of a taxonomy, requiring explicit understanding of the guiding theory. The review of interaction research suggests that categories constructed with a nursing situation in mind will be more likely to tap relevant dimensions of the interaction, and that application of category systems from other areas may not produce answers to nursing questions.We have indicated how, as nursing has moved from consideration of the technical, physical aspects of patient care to inclusion of the totality of patient experience, nursing research has had to include methodological models developed in the social sciences. The use of techniques of interaction analysis is an illustration of one situation in which nursing research has borrowed a method, and in some cases actual category systems for use in analyzing the nursing-patient interaction. We have commented on some of the difficulties of uncritical application of a social science method to nursing practice research, and have raised general issues in this regard.