In line with a dimensional approach to psychopathology, we examined whether psychosocial function and its relationship to neurocognition and clinical symptoms differ across schizophrenia and bipolar disorder subgroups with and without a history of affective or psychotic episodes. From the TOP study, a heterogeneous sample of individuals with schizophrenia spectrum disorders without (n = 60) and with a history of affective episodes (n = 54); individuals with bipolar spectrum disorders with (n = 64) and without a history of psychosis (n = 56) and healthy controls (n = 268) participated. Psychosocial functioning was measured with the Social Functioning Scale (self-rated) and the Global Assessment of Functioning Scale (clinician-rated), neurocognition with a comprehensive neuropsychological test battery, and symptoms with Inventory of Depressive Symptomatology, Young Mania Rating Scale, and Positive and Negative Syndrome Scale. Clinician-rated functioning was poorer in schizophrenia groups than in bipolar groups, but self-rated functioning was similar across all clinical groups and poorer than in controls. Neurocognition and current clinical symptoms were associated with psychosocial function in bivariate analyses, but current symptoms had a greater independent contribution to functioning than neurocognition across clinical groups in multivariate analyses. Despite differences in neurocognition and psychosocial function, groups showed the same pattern in prediction of functioning irrespective of DSM-IV or clinical definition.