Both hypertension and cognitive decline are common in old age. We sought to examine the effects of blood pressure (BP) on rates of cognitive decline in a longitudinal study of community-resident healthy old people. A total of 603 initially healthy old people aged over 69 years were visited at home. Subject's age, years of full-time education, Social Occupational Classification, health status and medication use were recorded. Sitting systolic and diastolic BP was measured, and the Mini-Mental State Examination (MMSE) and National Adult Reading Test (NART) administered. Follow-up was planned after 4 years: 69 subjects were dead, 15 were too unwell and 12 had moved away; 78 subjects either refused or failed to reply. Psychometric tests were administered to the remaining 429 (71.1%) after a median period of 4.20 years. Forty-two subjects had significant sensory impairment or interrupted testing. No significant differences in cognitive decline were found between those who had started medication (n = 163) and those remaining untreated (n = 224). Mean MMSE score change was 0.44 points (s.d. 2.07, P < 0.001). Entering all baseline variables into a stepwise regression analysis significant positive effects were found for initial MMSE score (beta = 0.50, P < 0.001), age (beta = 0.17, P < 0.001), systolic BP (beta = 0.16, P < 0.001) and period between testing (beta = 0.14, P = 0.004), and negative effect for NART-predicted IQ (beta = -0.16, P = 0.003).). We conclude that (1) older people exhibit faster age-associated cognitive decline as measured by MMSE; (2) people with higher NART-predicted IQs are relatively protected; (3) people with high systolic BPs are at greater risk of cognitive decline.