Studies on cognitive function in bipolar disorder have led to contrasting results and few data are available on affected subjects during the euthymic phase. In the present study we investigated the cognitive function of a cohort of bipolar (n=40) and schizophrenic (n=66) patients compared to healthy controls (n=64). Patients were evaluated in the outpatient setting over at least 3 months using a computerized version of Wisconsin Card Sorting Test. Schizophrenic patients showed the worst performance while that of the bipolar patients was somewhere between schizophrenic and controls. A discriminant analysis was able to classify correctly 60.59% of the subjects (schizophrenics 48.5%, bipolars 40%; healthy controls 85. 9%). The scores of the Wisconsin Card Sorting Test were entered into a principal component analysis, which yielded a 2-factor solution. Even in that analysis bipolar patients showed intermediate features in comparison with the other groups. These data indicate that bipolar patients have subtle neurocognitive deficits even after the resolution of an affective disorder. As well as observing quantitative differences between groups, the results show different dimensions of cognitive performance within groups suggesting that the deficit of euthymic bipolars could be a dishomogeneous entity, probably more heterogeneous than that in schizophrenia. Studies administering a more complete neuropsychological battery could further clarify the nature and meaning of the cognitive deficits in schizophrenia and bipolar disorder.