Background. This two-stage prevalence survey involved geographically delimited areas, four urban (Liverpool, Newcastle, Nottingham and Oxford) and two rural (Cambridgeshire and Gwynedd), including institutions.

Methods. Stratified random population samples of people in their 65th year and above, from Family Health Service Authorities were studied. The sample was stratified (65–74 years and [ges ]75) to provide equal numbers. In Liverpool equal numbers in 5 year age groups were taken. After an initial screening interview, approximately 20% were selected on the basis of age, AGECAT organicity confidence level and MMSE score to proceed to a detailed assessment interview from which the full AGECAT organicity confidence level could be derived.

Results. Major influences on MMSE were confirmed as age, sex, social class and educational level. Estimates of prevalence of AGECAT O3 and above for each centre and the entire sample according to age are given, based on 1991 Census population structure, and suggest that around half a million (543400) people in England and Wales would be defined as case level by this method. The five centres employing the same methodology showed no heterogeneity in prevalence.

Conclusions. Prevalence of cognitive impairment and dementia appear not to vary widely across the centres examined in this study, which provides stable estimates by age and sex for AGECAT O3 and above, and norms for MMSE. Using these estimates as an indication of the size of the population affected, around 550000 individuals in England and Wales would be expected to be suffering from dementia of mild or greater severity.