The effects of thioridazine on the performance of a titrating delayed matching-to-sample discrimination by four retarded adults was investigated. Trials began with the center of three response panels illuminated by one of three colors. The delay between depression of the center response panel and presentation of the two comparison stimuli on the side response panels varied according to the accuracy of the subjects' performance. The primary dependent variable was the limit of delay, defined as the longest delay at which the subject emitted four consecutive correct responses in a 30-minute session. The subjects' chronic doses of thioridazine were reduced systematically in a multiple baseline across-subjects design. For all of the subjects, the limit of delay increased after, and only after, reductions in the daily thioridazine dose had been implemented. Results indicated that the withdrawal of chronically administered thioridazine resulted in increased accuracy in a delayed matching-to-sample task, suggesting strongly that the drug impairs performance of this discrimination.