Forty-three delinquent boys were randomly assigned to double-blind treatment with diphenylhydantoin sodium, methylphenidate, or placebo for two weeks. Ratings of symptoms by cottage parents and teachers, a measure of frustration, the Porteus Mazes, and an interview were used to assess the effects of the treatments. None of the measures showed effects attributable to the drugs, and subjective reports tended to show negative effects of the active drugs. The danger of random assignment studies was demonstrated, however, by the fact that the more disturbed children were assigned by chance to the placebo group. The lack of severe symptomatology, the short period of treatment, and the heterogeneous nature of the sample were considered as possible explanations of the apparent lack of beneficial effects of the two drugs.