We propose that depressive deficits in remembering are revealed in tasks that allow the spontaneous use of strategies; tasks that bypass or direct the use of strategies should not produce depressive deficits. College students received depressive-or neutral-mood inductions after answering questions worded to reflect homophones' less common meaning. After the inductions, subjects spelled old and new homophones and showed no effect of the depressive inductions on unaware memory for the old homophones. Subsequent tests of recognition did, however, reveal differences according to the induced mood or the presence of naturally occurring depression (in Experiment 3). The differences, evidence of nondepressed subjects' use of strategies, tended to disappear when all subjects were provided with strategies for spelling or recognition. The results indicate that depressives experience deficits in cognitive initiative. We review the literature on depressive memory from this perspective.