The distinction between bipolar disorder and schizophrenia customarily follows examination of the clinical symptomatology and course of illness. The presence of cognitive impairment has been held to be uncommon in bipolar disorder and more likely in schizophrenia. This study explored neuropsychological function in 30 ambulatory outpatients with a DSM-III-R diagnosis of bipolar affective disorder (all of whom had been psychotic during manic episodes), comparing their performance with that of controls. These bipolar patients proved to have significant levels of diffusely represented cognitive impairment when compared with controls. Further, the degree of impairment was significantly correlated with reduction in midsagittal areas of brain structures measured on magnetic resonance imaging scans. The implications of these findings in relation to bipolar disorder are discussed.