39 children in a psychiatric hospital were tested on 2 types of learning tasks and several performance tasks before, during, and after a 4-week period on chlorpromazine or placebo, the types of medication being assigned randomly in a double-blind design. Medication was controlled according to clinical criteria. Paired-associate learning was less efficient under chlorpromazine, especially on later learning trials and among initially slower learning Ss. Serial learning and tapping rate showed less consistent trends. Porteus Maze Mental Age scores declined under chlorpromazine, while Q scores on the same test tended to improve. No drug effect was noted on remote or immediate memory. The results were tentatively interpreted in terms of impairment of attention span by chlorpromazine.