Depressive realism suggests that depressed individuals make more accurate judgments of control than their nondepressed counterparts. However, most studies demonstrating this phenomenon were conducted in nonclinical samples. In this study, psychiatric patients who met criteria for major depressive disorder underestimated control in a contingent situation and were consistently more negative in their judgments than were nondepressed controls. Depressed patients were less likely than their nondepressed counterparts to overestimate control in a noncontingent situation, but largely because they perceived receiving less reinforcement. Depressed patients were no more likely to use the appropriate logical heuristic to generate their judgments of control than their nondepressed counterparts and each appeared to rely on different primitive heuristics. Depressed patients were consistently more negative than their nondepressed counterparts and when they did appear to be more " accurate" in their judgments of control (as in the noncontingent situation) it was largely because they applied the wrong heuristic to less accurate information. These findings do not support the notion of depressive realism and suggest that depressed patients distort their judgments in a characteristically negative fashion. © 2009 Elsevier Ltd.