This article contrasts ‘mainstream’ narrative analysis, and the study of researcher-elicited narrative accounts, with conversation analysis and the study of naturally occurring narratives-in-interaction. This analysis extends previous conversation analytic and discursive psychological work on storytelling (i.e., how stories get embedded in sequences of talk; the actions storytelling does), by focusing on the location and function of speakers’ story formulations and orientations to narrative (e.g. “I think we should start at the beginning”, “You want the full story, or…?”, “there’s always two sides to every story”). Rather than treating such ‘meta-formulations’ as partial expressions of a general folk theory of narrative, we examine their action-orientation and the way they are shaped for the occasions of their production; how members’ commonsense notions of stories are displayed in the interactional contexts in which they are put to use. The argument is illustrated by a range of brief examples from mundane conversation, police interrogation, and neighbour dispute mediation.