To the Editor.
—The National Institute of Mental Health Diagnostic Interview Schedule (DIS) was described by its originators, Robins et al, in the April 1981 issue of the Archives (1981;38: 381-389) and was subject to further discussion in the correspondence from Endicott and from Heizer and Robins in the November 1981 issue (1981;38: 1300-1301). We have been using the DIS ourselves in a study of the recognition and management of depressive disorders in the psychiatric and primary medical care sectors, and have the following observations.In a completely structured interview like the DIS, we do not expect reliability to be low, and the initial data from St Louis support this. We are, however, less confident of its validity. As Endicott pointed out in her letter, agreement between computer-derived diagnoses indicat 3 high reliability but not necessarily validity because there is no external criterion of validity, such as a clinical diagnosis.