The study attempted to isolate some of the environmental conditions that controlled the self-destructive behavior of three severely retarded and psychotic children. In the extinction study subjects were placed in a room where they were allowed to hurt themselves, isolated from interpersonal contact. They eventually ceased to hurt themselves in that situation, the rate of self-destruction falling gradually over successive days. In the punishment study, subjects were administered painful electric shock contingent on the self-destructive behavior. (1) The self-destructive behavior was immediately suppressed. (2) The behavior recurred when shock was removed. (3) The suppression was selective, both across physical locales and interpersonal situations, as a function of the presence of shock. (4) Generalized effects on other, non-shock behaviors, appeared in a clinically desirable direction. Finally, a study was reported where self-destructive behavior increased when certain social attentions were given contingent upon that behavior.