THE ADVANCE of any science requires a widely shared and explicit definition of the phenomena which it studies. For effective clinical intervention and research psychiatry needs a reliable classification system to facilitate communication among its professionals. At present there is a commonly held impression, backed by a number of studies,1-8 that the current psychiatric categories do not adequately meet this criterion of reliability. Some have advocated a totally new nosologic system, while others have reacted to this situation by downgrading the importance of diagnostic labels. Even if a new diagnostic system were devised and judged to be adequate, it would take years for it to gain widespread acceptance and use. Such a strategy also implies that the accumulated observations and traditions of clinicians for the past century have little value. While attempts to construct better diagnostic systems are not to be discouraged, an alternate plan might