The well regarded Section of the World Psychiatric Association, on Epidemiology and Public Health, held its biennial meeting at Johns Hopkins University, Baltimore in June 2001. A new feature of the meeting was the ‘work group’, an opportunity to debate topical but unresolved issues. There was standing room only for a discussion on why the differences between self-report and clinician-rated measures of psychopathology matter (Brugha et al. 1999a). This is one of several topics increasingly debated in the context of the seemingly unremitting disparities between epidemiological estimates of the prevalence of psychiatric disorders (Regier et al. 1998). Recent comparisons of clinician and lay (self-report) measures suggest that dichotomous diagnostic categories cannot be measured reliably in large scale community surveys against a clinical standard measure (Brugha, et al. 1999b, 2001; Eaton et al. 2000).