This essay explores the question of relevant scale: which of the many potentially relevant processes – from interactional through local through global, from nearly instantaneous through those emergent over months, years or centuries – in fact contributes to social identification in any given case, and how do these heterogeneous processes interrelate? Contemporary answers to this question have moved beyond the détente of the “micro-macro dialectic,” in which purportedly homogeneous “macro” processes constrain events and actions, while being simultaneously constituted by “micro” events and actions. We review contemporary work on these issues, with particular reference to the use of language in social identification, and we argue that an adequate account must go beyond “micro” and “macro.” We illustrate our argument with data from a seven-year ethnographic project in an American town that has received thousands of Mexican immigrants over the past decade, focusing on two types of narratives that residents tell about immigrants: stories about “payday muggings” in which immigrants are victimized, and stories about the town's historical trajectory and immigrants' role in it. These narratives emerge and move across different scales, and they are an important resource for residents as they socially identify themselves and others.