The revision of the DSM scheme has highlighted several difficulties with classification: adapting criteria to different ages and genders; multiplicity of diagnoses in individual children (leading to a mistaken impression of ‘comorbidity’); complexity of the scheme making problems for nonspecialist clinicians; the use of unvalidated categories; and the inclusion of functional impairment as a criterion for several disorders.

In this paper, DSM5's approaches to solving some of these difficulties are illustrated, with reference to specific conditions arising in childhood; although at the time of writing full criteria are not available. An overarching formulation of disorders has been introduced, and “disorders arising in childhood” no longer form a separate group in view of the extensive evidence for the early origins of many psychiatric disorders. “Pediatric bipolar disorder” has become overused, and a new category of “disruptive mood dysregulation” introduced. ADHD subtypes have not been validated and are downgraded to “presentations”; and “restrictive-inattentive” disorders have not (yet) earned their place. ADHD criteria in adult life have been relaxed a little. The “autism spectrum” has replaced subsyndromes, and proposes two dimensions for description. Callous-unemotional states in conduct disorder have received attention; as has the description of syndromes involving self-harm. The remaining challenges for clinicians and researchers will be described.