Two experiments examined the contribution of reporting biases to mood-congruent recall patterns and diminished levels of recall frequently associated with depressed mood states. In Experiment 1, participants classified as dysphoric (n = 14) or nondepressed (n = 21) on the basis of scores on the Beck Depression Inventory and the Profile of Mood States made self-referential judgements regarding a series of affectively valenced words. Subsequently they were given an unexpected forcedrecall test, which encouraged guessing to meet the output requirement (i.e. 40 responses) of the test. Nondepressed subjects confidently reported more positive words than dysphoric subjects, but the latter produced significantly more correct guesses of words that were positively valenced. Similar findings were obtained in Experiment 2, in which dysphoric (n = 40) and nondepressed subjects (n = 40) performed both self-referent and orthographic judgements of affectively valenced words, followed by either a free- or forced-recall test. The findings suggest that positive and negative trait words were adequately encoded in memory, but, consistent with cognitive theories of depression, their accessibility to retrieval was differentially limited. In addition, however, the results implicate an important contribution of diminished motivation and/or conservative report criterion in the manifestation of depression-related biases and deficits in recall.