The effects of haloperidol and a mild punishment on the severe self-injurious behavior and several collateral behaviors of a 17-year-old profoundly retarded male were assessed. A 12-month analysis using a withdrawal design suggested that neither the medication nor the behavioral intervention alone was effective in significantly reducing the frequency of self-injurious behavior. When combined, however, these interventions produced dramatic reductions in the subject's self-injurious behavior. The haloperidol may have acted as a "setting event" for the successful use of the punishment. Suppression of this behavior was maintained at 6 months and 1 year following the end of the analysis. The collateral behaviors were differentially affected by the behavioral and pharmacological interventions. Time spent in bed and the appearance of drooling increased with the introduction of the haloperidol, while percent correct on a fine-motor task increased only when the interventions were applied simultaneously. The results point out the importance of a careful behavioral analysis for both pharmacological and behavioral interventions and their possible combined actions.