Three neuropsychological tests (Rey's auditory verbal learning test, word fluency and signal detection test for words) were administered to 36 depressed patients (medicated and nonmedicated) and 26 controls and compared to scale scores for depression severity and psychomotor retardation to examine how retardation was related to cognitive performance. As expected, compared to controls depressives exhibited non-specific ''cognitive inefficiency'', that is, consistently deficient performance in all cognitive tasks. Results on all three tests were correlated with one other and with depressive severity (measured by the Montgomery & Asberg Depression Rating Scale, MADRS). One notable exception, however, was the score for commission errors (''false alarms'') in the signal detection test, which correlated negatively with psychomotor retardation (measured by a subscale of the Salpetriere Retardation Rating Scale, SRRS) but not at all with depressive severity. Lack of commission errors thus seemed to index a dimension of retardation of ideation that seemed distinct from the non-specific cognitive inefficiency dimension. Conversely, omission errors in the same test strongly correlated with the other two cognitive tests and with depressive severity but not with psychomotor retardation. Cognitive performance in depressives might thus be explainable in terms of two overlapping dimensions of depressive pathology (global cognitive impairment vs. specific effect of retardation). Further studies with non-medicated patients are needed to determine to what extent these findings may be due to medication effects.