Operant conditioning techniques were applied to the study of stereotyped body-rocking movements. Six mentally retarded children were tested individually in a fully automated situation under the following conditions: (a) free operant baseline; (b) intense auditory stimulation; (c) variable-interval (VI) and fixed-ratio (FR) reinforcement schedules; (d) differential reinforcement of low rate (DRL); (e) differential reinforcement of other behaviors (DRO); and (f) noncontingent reinforcement. The results indicated: (a) intense auditory stimulation increased the rate of rocking movements in four of six subjects; (b) VI and FR schedules generated curves that are representative of these schedules; (c) DRL schedules decelerated high-rate rocking; (d) DRO schedules eliminated high-rate rocking; (e) noncontigent reinforcement increased variability and resulted in decreased rate of responding. In conclusion, the data support the hypothesis that body-rocking movements in the mentally retarded are an operant and reflect the established characteristics of reinforcement schedules and operant controls.