This article follows Blaxter’s foundational call for a sociology of diagnosis that addresses the dual aspects of diagnosis‐as‐category and diagnosis‐as‐process. Drawing on video recordings from an autism clinic, we show how the process of attaching the diagnosis to a child involves interactions between clinicians, parents and children, and that in the course of such interactions, a diagnostic category officially defined in terms of deficits can instead be formulated in terms of valuable social and cognitive differences. More specifically, we show that the child’s age is crucial for how clinicians formulate the diagnosis: with younger children, clinicians treat autism exclusively as a deficit to be remedied, whereas with older children, clinicians may treat autism either as a deficit or as a social‐cognitive difference. Further, because older children are often co‐recipients of diagnostic news, we find that clinicians carefully manage the implications such news may have for their self/identity. Finally, we suggest that (1) the equation of a diagnostic category with dysfunction is achieved in interaction; (2) the meaning of a diagnosis may vary with characteristics of its recipients; and (3) that meaning can be worked up by clinicians and recipients in ways that centre difference rather than deficit.