Thirty-eight boys, referred for psychiatric treatment because of serious problems of behaviour, underwent a double-blind, placebo-controlled, crossover trial of methylphenidate and placebo. Methylphenidate was an effective treatment over a 3-week period. A good response to methylphenidate was predicted by higher levels of inattentive and restless behaviour, impaired performance on tests of attention, clumsiness, younger age and by the absence of symptoms of overt emotional disorder. DSM-III and ICD-9 diagnoses of 'hyperactivity' were not good predictors. The results support the validity of a construct of hyperactivity in describing childhood psychopathology, but emphasize the need for a refinement of diagnostic criteria.