Vignettes representing mentally ill people were presented to about 2000 randomly selected residents in Bromsgrove, served by a mental hospital, and Malvern, served by a community-based psychiatric service. They were asked about the likely cause of the condition, what action they would take, and what agencies were most likely to help. In Malvern, residents seemed more enterprising in involving various agencies and more tolerant. It is possible to derive simple 'action scores' as an indicator of such tolerance. Although there were significant demographic differences between subgroups, overall identification of vignette subjects as mentally ill was surprisingly low, and so was knowledge both of community psychiatric nurses as an agency, and of the location of dispersed treatment facilities in both areas.