BACKGROUND: Deteriorating cognitive function in late life substantially increases the risk for dementia, for other non-cognitive morbidity, for dependency, and early death. AIMS: To identify early predictors of late-life cognitive outcome. METHOD: Cognitive function, premorbid IQ, and cardiovascular risk exposure were recorded on 1083 subjects on entry to a hypertension treatment trial in 1983-1984. We followed up this cohort 9-12 years later to assess cognitive function with the Mini-Mental State Examination (MMSE), to update exposure status, and to obtain genomic material. Multivariate analysis was used to identify independent baseline predictors of cognitive outcome 9-12 years later. RESULTS: We followed up 387 subjects (58.6% of survivors). After adjusting for baseline cognition, poorer cognitive outcome was found to be independently associated with a family history of dementia, increasing age, less decline in systolic blood-pressure, lower premorbid IQ (rather than limited education), and abstinence from alcohol. CONCLUSIONS: Reduction in systolic blood pressure (among hypertensives) and moderate alcohol intake could protect against cognitive deterioration in late life.