Dysphoric and nondysphoric students (48 women and 24 men) participated in an experiment that was designed to separate automatic and controlled uses of memory in a modified recognition paradigm. First, they judged the relation of target words to paired words. Later they made recognition decisions on target items alone or in the context of the original paired item. The use of L.L. Jacoby's (1991) process dissociation procedure revealed depressive deficits in estimates of recollection but not in estimates of familiarity. The paired test improved recollection for all subjects and showed a trend in the direction of increased familiarity. These outcomes support approaches to depressive cognition that emphasize impaired cognitive control.