Outcome studies of schizoaffective disorder have taught us much about the long-term consequences of the syndrome, and they have provided some indication of the potential usefulness of maintaining "schizoaffective disorder" as a diagnostic category separate from schizophrenia and major affective disorder. In a review of outcome studies that compared schizoaffective patients to schizophrenic or affective patients, we found consistent results despite wide variations in diagnostic criteria, length of followup, and demographic characteristics. Global measures of outcome show that schizophrenic patients are more impaired than schizoaffective patients, who in turn are more impaired than affective patients. However, studies of specific outcome domains such as symptomatology, social functioning, and occupational functioning indicate that schizoaffective disorder is heterogeneous and that subtyping by polarity (e.g., schizoaffective-manic vs. schizoaffective-depressed) accounts for some of this variance. The consistency of these findings in the face of methodological variability suggests that it would be premature to classify schizoaffective patients with schizophrenia or affective disorder, but also that strict diagnostic criteria for schizoaffective disorder are at best preliminary and need to be thoroughly validated.