This study was undertaken to determine the utility of the Minnesota Multiphasic Personality Inventory (MMPI) for prediction of psychiatric diagnoses. The recent development of operationally defined psychodiagnostic criteria facilitated tests of this standard personality inventory for prediction of types of mental disorder with specific inclusion and exclusuion criteria, a task for which the MMPI was intended;Three hundred seven male psychiatric patients were diagnostically classified using the Research Diagnostic Criteria (RDC). Their membership in psychopathological groups was predicted from MMPI clinical and content scales via multiple discriminant function analyses, Goldberg's linear rules, and the Meehl-Dahlstrom configural rules. Results confirmed that the MMPI clinical scales were significant and accurate predictors of specific and molar diagnoses. Multiple discriminant function prediction was more accurate than that of linear or configural rules. Cross-validation demonstrated strong stability for the multivariate prediction of RDC categories. The MMPI content scales yielded less robust, though significant, prediction than clinical scales. The relationships between these results and those of previous investigations were discussed; suggestions for extension and replication of these results were offered.