Six different models for assessing psychiatric impairment were applied to 188 men who had been studied prospectively for half a century. The six models were categorical DSM-III diagnosis by a senior psychiatrist; objective evidence of impairment in working and loving; relative maturity of ego mechanisms of defense; the Health Sickness Rating Scale; the Global Assessment Scale (Axis V of DSM-III-R); and a "research" scale designed to provide a consensual definition of "caseness." As a test of predictive validity of our impairment measures, assessments before the age of 50 years were used to predict psychosocial adjustment and physical health assessed 15 years later. All six models were highly intercorrelated and proved equally effective in predicting future impairment. Procedures to establish construct validity suggested that the presence of a DSM-III diagnosis was a valid categorical definition of case, and that Axis V of DSM-III-R was a valid dimensional measure of impairment.