Subjects with depression exhibit deficits in prefrontal function. We posited that as a result, in a supraspan memory test, they would be impaired in their ability to inhibit recall of irrelevant words, and because of consequent overload of working and episodic memory capacity, would be impaired in their ability to recall relevant words. We tested this hypothesis in 30 inpatients and outpatients with a diagnosis of major depressive disorder and 30 controls subjects using a form of the Directed Forgetting Paradigm using exclusively neutral words. The depressed subjects did exhibit deficits in prefrontal function. All subjects were given four lists of 24 items each, in which half the words were followed by the instruction “remember” and half by the instruction “forget.” Our hypothesis found support in a significant group by item type interaction effect exhibited when subjects were instructed to recall only those items followed by the “remember” instruction: depressed subjects recalled relatively more words to be forgotten and relatively fewer words to be remembered. A control experiment suggested that these results could not be accounted for by a differential effect of depression on memory encoding. (JINS, 2008, 14, 895–899.)