OBJECTIVE: The author examined the validity of schizophreniform disorder as defined in DSM-III and DSM-III-R by critically reviewing the current literature. METHOD: Thirty publications that met specific inclusion criteria were identified and reviewed. To organize information relevant to the question of validity, the data were divided into three general classes of validators: antecedent, concurrent, and predictive. When possible, meta-analyses of specific validators were performed according to the summary odds ratio method of Mantel- Haenszel. RESULTS: The findings of this review failed to provide support for the validity of schizophreniform disorder, as it is currently defined, as a distinct diagnostic entity. In addition, the review did not provide strong support for schizophreniform disorder as representing a subtype of schizophrenia or affective illness. Instead, the data suggest that the current diagnostic criteria for schizophreniform disorder identify a heterogeneous group of patients with new-onset schizophrenia, schizoaffective disorder, and atypical affective disorder and a small subgroup with a remitting nonaffective psychosis. CONCLUSIONS: This review suggests that the diagnosis of schizophreniform disorder has limited clinical utility. Patients meeting criteria for schizophreniform disorder should instead be diagnosed as having psychotic disorder not otherwise specified until additional clinical information (e.g., course of illness) becomes available. Future research will require prospective longitudinal follow- up of patients with new-onset affective and nonaffective psychoses to identify the subgroup with "true" schizophreniform disorder, i.e., a remitting nonaffective psychosis.