The validity of DIS-CM (Chinese modified version of Diagnostic Interview Schedule) was examined by analyzing lifetime prevalence of each age group, age at onset, and recency of illness. The magnitude of the discrepancy between the empirical and the estimated data was adopted as the criterion of the degree of deviation. Manic and depressive episodes were found to be seriously underestimated. Schizophrenic, panic, phobic and antisocial personality disorders were probably underestimated. The observed lifetime prevalence figures with DIS-CM are rather conservative for these disorders. The prevalence data of a recent 1-year period are less biased and more reliable. The nature of the disease, recall effect, active or passive psychological resistance, mortality, and uncooperative attitude are considered factors that induce an underestimate of lifetime prevalence.