Recent decades have witnessed a dramatic upsurge in the prevalence of autism spectrum disorder (ASD). As researchers have investigated the responsible sociohistorical conditions, they have neglected how clinicians determine the diagnosis in local encounters in the first place. Articulating a position "between Foucault and Goffman," we ask how the interaction order of the clinic articulates with larger-scale historical forces affecting the definition and distribution of ASD. First, we show how the diagnostic process has a narrative structure. Second, case data from three decades show how narrative practices accommodate to different periods in the history of the disorder, including changing diagnostic nomenclatures. Third, we show how two different forms of abstraction-Type A, which is categorical, and Type B, which is concrete and particular-inhabit the diagnostic process. Our analysis contributes to the sociology of autism, the sociology of diagnosis, the sociology of abstraction, and social theory.