Because of such validity-research deficits and the ceiling on agreement between instruments imposed by less-than-perfect reliability characteristics of each instrument, it is not appropriate to assume that the semistructured clinician interview is more valid than the epidemiologic interview. The Baltimore ECA site is uniquely situated to address this issue by comparing the outcome of subjects identified with current depression in the 1982 clinical reappraisal interview with those identified by the DIS at the same time to see if the 13-year follow-up is similar to that found over 16 years by Murphy et al. Where do we go from here in improving our diagnostic criteria for DSM-V, constructing better diagnostic instruments, and conducting the next generation of epidemiologic studies? Certainly the categorical diagnostic criteria themselves, without a dimensional symptom level, are never used in clinical treatment trials. Hence the "clinical significance" criteria of significant distress or disability added to DSM-IV should be further refined, with the possible addition of "staging" of disorders. The objective would be to provide a better indication of treatment need and clinical prognosis as in current cancer diagnostic assessments. For epidemiologic studies, the addition of symptom scales and disability assessments to the traditional categorical diagnoses should be helpful in developing community measures of treatment need. Different methods of assessment may be useful for diagnoses in which an impaired perception of reality occurs, such as schizophrenia. With some of these adjustments, it should be feasible to "count" those with clinically significant diagnoses in the community, and thus improve the validity and clinical utility of our diagnoses for predicting clinical course and responsiveness to specific treatments.