Response-contingent electric shock was evaluated in a laboratory and criterion situation before using it to eliminate infrequent and unpredictable biting and destructive acts from the repertoire of a retarded adolescent. Although the initial results were dramatic, the effects of shock were highly discriminated and the therapeutic objectives were not accomplished. The study demonstrates the power and some limitations of shock punishment and is presented so that others will be better prepared for problems that may arise before the goals of a program involving shock punishment will be realized.Response-contingent aversive events have been used by several investigators to decelerate problem behavior of children (Bucher and Lovaas, 1967;Lovaas, Freitag, Gold, and Kassorla, 1965;Risley, 1968;Tate and Baroff, 1966;Wolf, Risley, and Mees, 1964; Wolf, Risley, Johnston, Harris, and Allen, 1967). Punishment, together with positive reinforcement of other behavior, was applied to protect the patient from himself or to protect others. The results in each study were dramatic and no undesirable "side effects" were observed. But these investigators also found that the results were situation-or response-specific, a common finding in the punishment literature. (See reviews by Azrin and Holz, 1966;Solomon, 1964.) How to maintain improvements and to achieve generalized suppression is a practical problem of formidable dimensions. (Feldman, 1966;Feldman and MacCulloch, 1965) and "office treatment" of behavior assumes transfer to the patient's natural environment, an assumption which hardly conforms to the available data.The subject used in the present study presented an opportunity to study generalization effects because the nature of the problem behavior required an indirect approach. The target behavior, biting other children, did not occur in the presence of observers and occurred very infrequently (eight times in the 125 days preceding this study). The patient was continuously restrained in isolation and release was contingent upon a treatment program that entailed minimal danger and showed promise of success.The study continued for 16 months and included treatment and recording of results in a laboratory a criterion situation, and the patient's ward. Since the patient did engage in a variety of other undesirable acts, such as destroying furnishings and tearing clothing when observers were present, the approach taken was to threaten shock for such responses, administer shock response-contingently when the threat was insufficient, and repeat the procedure with other responses. In other words, controlled conditions were arranged to establish that shock would suppress behavior and to establish warnings-"Don't!", "No!"-as generalized conditioned aversive stimuli with the 201 NUMBER 3 (FALL 1968) 1968, 1, 201-21 1