The centrality of emotion in thought and action is increasingly recognized in the human sciences, though basic questions of definition and scope remain unresolved. Where do emotions begin and end? How should we identify and analyse them? How should we write about them? Ethnographic fieldwork, as pioneered by Malinowski, offers powerful insights into the place of emotion in social life; but emotions are peculiarly difficult to capture in the generalizing format of case study and ethnographic summary. In this article I argue that semantic, structural, and discourse‐based approaches tend to miss what is most important – what counts for the persons concerned and therefore what makes the emotion. I review the conceptual and methodological issues and conclude that only a narrative approach can capture both the particularity and the temporal dimension of emotion, restoring verisimilitude and fidelity to experience.