In a previous report records were studied of 256 in-patients who were discharged as undiagnosed, i.e. in-patients who at the time of discharge did not have a definable psychiatric illness (Welner, Liss, Robins and Richardson, 1972). In that study it was shown that when rigorous criteria for psychiatric research (Feighner, Robins, Guze, Woodruff, Winokur and Munoz, 1972) were used 68 per cent of these patients met the criteria for an established psychiatric disorder. It was concluded that: (1) The chart review diagnoses for a population of undiagnosed patients consisted of a variety of established psychiatric disorders and the population was not homogeneous. (The chart review diagnosis is a diagnosis obtained by review of the patients' hospital records and evaluating the information by using diagnostic criteria for psychiatric disorders.) (2) The most efficient way to arrive at a diagnosis was by structured rather than conventional narrative interview. This study is a follow-up study of these patients and attempts to evaluate the validity of the chart review diagnosis. A concordance between the chart review diagnosis and follow-up diagnosis supports the above conclusions. The follow-up study also served to establish diagnosis in patients who had too few symptoms initially to meet the criteria for a diagnosis.