An experiment is reported in which a depressed and a control group were tested on free recall, cued recall and recognition memory for a prose passage. As expected from previous work the depressives tended to show less impairment on recognition than on free recall. However, contrary to what some theories would predict, cued recall performance was no better than free recall. The implications of this finding for the nature of the depressive memory deficit for neutral materials are discussed. It seems that neither the amount of verbal output required, nor the need to generate retrieval cues, are critical factors.