This article identifies what it calls a 'performative turn' in historical studies, as in the humanities more generally. It traces the rise of the notion of performance out of the dramaturgical model of the 1940s and 1950s (associated with Kenneth Burke, Erving Goffman and Victor Turner), linking the new idea (developed by John Austin, for instance, and Pierre Bourdieu) to the rise of 'postmodernity'. Turning to historical studies, the article analyses the role played by the concept of performance in recent studies of ritual, festivals, identity, gender, and even emotions, architecture and knowledge, noting the shift from the assumption of social or cultural fixity to that of fluidity, from scripts to improvisations, from mentalities to the habitus. The new approach has generated problems, notably the overreaction against the idea of social constraints, the danger of circularity and the over-extension of the central notion of performance. A stronger and a weaker sense cl 'performance' need to be distinguished. Despite these problems, the performative approach has foregrounded important and neglected issues, notably that of differences between cultural or social domains. The article concludes by considering the methodological consequences of the approach, in particular its implications for source criticism and historical explanation.