In the absence of selective agonists and antagonists able to discriminate between individual members of the D1-like and D2-like families of dopamine receptor subtypes, functional parcellation has remained problematic. 'Knockout' of these subtypes by targeted gene deletion offers a new approach to evaluating their roles in the regulation of behaviour. Like any new technique, 'knockout' has associated with it a number of methodological limitations that are now being addressed in a systematic manner. Studies on the phenotype of D1(A/1), D(1B/5), D2, D3 and D4 'knockouts' at the level of spontaneous and agonist/antagonist-induced behaviour are reviewed, in terms of methodological issues, neuronal implications and potential clinical relevance. Dopamine receptor subtype 'knockout' is a nascent technology that is now beginning to fulfil its potential. It is being complemented by more systematic phenotypic characterisation at the level of behaviour and additional, molecular biologically-based approaches.