REGIER ET al1 summarize discrepancies in prevalence rates between the Epidemiological Catchment Area (ECA) study and the National Comorbidity Survey (NCS) and provide a useful overview of plausible explanations. I will extend their comments by emphasizing 3 inherent limitations to defining clinical cases in epidemiological studies: (1) the definition of mental disorder in DSM-IV2 fails to provide a clear boundary between psychopathology and normality; (2) the concepts "clinical significance" and "medical necessity" are difficult to operationalize and to assess reliably; and (3) lay interviewers do not have the experience necessary to judge clinical significance.