To estimate 10 year decline in cognitive function from longitudinal data in a middle aged cohort and to examine whether age cohorts can be compared with cross sectional data to infer the effect of age on cognitive decline.