This paper reports on the predictive validity of the physical disorders axis (axis III) of the DSM multiaxial diagnostic system at 3-year follow-up. A total of 515 general psychiatric patients were assessed with a semistructured procedure that covers all DSM-III diagnoses and axes, and were subsequently followed up for 3 years. Outcome was assessed with several measures of adaptive functioning. Baseline axis III was analyzed according to a) presence of any physical disorder, b) the number of these, c) presence of major chronic physical disorders (MCPD), and d) the number of these. Prediction of impairment in functioning (Strauss-Carpenter Scale), derived from baseline axis III, ranged from a correlation coefficient of .18 when expressed as the presence of any physical disorder to .35 when represented by the number of MCPD. Furthermore, within patients with specific psychiatric disorders, it was found that number of MCPD reached a predictive validity of .55 for patients with dysthymic disorders, .44 for those with anxiety disorders, and .41 for those with major depression. Comparative multiple regression analyses, controlling for demographic and clinical variables, showed that the number of MCPD at baseline was the most important predictor of functioning outcome among patients with dysthymic disorders and major depression. The number of MCPD experienced by general psychiatric patients seems to be an important predictor of future functioning, particularly for patients with certain psychiatric disorders. This points out the importance of considering the relationship between psychiatric and MCPD when conducting systematic clinical assessments towards the prediction of course and outcome.