OBJECTIVE: This study was designed to ascertain the degree and specificity of cognitive impairments in patients with schizophrenia and patients with affective disorders. METHOD: Cognitive function was assessed with a neuropsychological test battery in consecutively admitted patients with schizophrenia (N = 57), unipolar depression (N = 29), and bipolar disorder (N = 16). RESULTS: The performance of the schizophrenic group was significantly below that of the groups with affective disorders on measures of attention and psychomotor speed, verbal and visual memory, and problem solving and abstraction. IQ was lower in the schizophrenic group and appeared to have deteriorated from a normal premorbid level that was not different from that of the affective disorder groups, as determined by the Wide Range Achievement Test--Revised reading test, a putative measure of premorbid intelligence. When IQ was controlled, differences between the groups in problem solving and visual memory remained. Psychiatric symptoms had a larger impact on test performance in the affective disorder groups than in the schizophrenic group. CONCLUSIONS: These results suggest that patients with schizophrenia perform systematically worse on cognitive measures than patients with affective disorders, which is consistent with their generally poorer outcome. The results also indicate that schizophrenia and affective disorders are qualitatively distinguishable in neuropsychological terms, given differences in apparent intellectual deterioration, profiles of cognitive impairment, and associations between cognitive performance and psychopathology.