All diagnosis depends on communication between doctors and patients. This is especially so with behavioural disorders such as autism, where structured interactions involving clinicians and children (e.g. standardised tests) play a key role in diagnosing the condition. Although such interactions are collaborative, we find that when reporting test results, clinicians, following administrative protocols, routinely gloss over the embodied interactions constitutive of testing, such that autism is predicated as an inherent feature of the child. In ethnomethodological terms, this is related to the way that "accounts" (Garfinkel 1967), including diagnoses, are reflexively related to the taken-for-granted practices that make them objectively reportable in prevailing professional terms. These practices include how the clinicians themselves interact with children they examine, with other professionals, and with the instruments used to test a child. Examining video footage of a multi-stage autism evaluation, along with the medical report rendering the child's diagnosis, we show how reporting practices, while addressing the administrative features of standardised testing and diagnosis, can also be examined for their grounding in an environment of tacit matters usually unavailable for inspection. We conclude by asking whether, and how, oral and written reports might re-situate children in the concreteness of their social environments. © 2019 Foundation for the Sociology of Health & Illness.