Narratives told in interview have become a central tool of data collection and analysis in a variety of disciplines within the social sciences. However, many researchers, particularly those who embrace a conversational analytic or ethnomethodological approach (see among others Schegloff, 1997; Goodwin, 1997), regard them as artificial and oppose them to naturally occurring stories, which they see as much richer and interesting sources of data and analysis. In this paper, I argue that the criticism against interview narratives has been justified by the lack of attention that many narrative analysts have shown towards the interview as a truly interactional context. However, I also point to some shortcomings that derive from this opposition between naturally occurring and interview narratives and to an alternative framework in which the stress is not on the kind of narrative data used for the analysis, but rather on the kind of narrative analysis that should be adopted. I argue that our methodologies of analysis cannot fail to take into account the way narratives shape and are shaped by the different contexts in which they are embedded and propose the study of narrative genres as a way of looking at the reciprocal influence of narratives and story-telling contexts. I illustrate this point looking at accounts as a genre.