Twenty-four psychiatric staff, 24 clinically depressed inpatients, and 24 nondepresssed schizophrenic patients at a state psychiatric facility completed five tasks under either reward or punishment conditions. Each task consisted of 30 trials of pressing or not pressing a button to make a light appear. Monetary reinforcement was contingent on light onset for the final ten trials of each task. Cash incentives for judgment of control accuracy were added for Tasks 3, 4, and 5. Cognitive functioning was evaluated on each task by measuring expectancy, judgment of control, evaluation of performance, and attribution. Mood and self- esteem were measured before and after the procedure. No significant differences were observed across mood groups for expectancy of control or judgment of control accuracy. Subject groups also did not differ in the attributions they made or in how successful they judged their performances to be. They set realistic, attainable criteria for success which were consistent with relevant conditional probabilities. Subjects in reward gave themselves more credit for task performance than subjects in punishment gave themselves blame for comparable performances. Punishment subjects demonstrated more stable, external attributions than those in reward. Across tasks, subjects overestimated when actual control was low and underestimated when actual control …