The relationship between blood pressure and health in old age is complex and influenced by socio-economic factors. The Healthy Old People in Edinburgh cohort were initially disease-free and untreated, providing a sample in which directionality in this relationship could be examined. Subjects' health status, medication use and blood pressure was ascertained at baseline, after 4 years, and again after 7 years. Socio-demographic and socio-economic data were also collected. A total of 603 subjects were seen at baseline, 429 at 4 years and 301 at 7 years; complete blood pressure data were available for 294. Mean blood pressures were 157/85 mm Hg, 159/87 mm Hg and 162/86 mm Hg at baseline, 4 years and 7 years respectively. When subjects with diagnosed hypertension were excluded, the presence of disease (P = 0.009) and medication use (P = 0.047) at 7 years were associated with a relative reduction in blood pressure over time. For these subjects disease was predicted by deprivation index of residential area (OR 1.24, 95% CI 1.10-1.40 per Carstairs unit) and occupational group (OR 0.85, 95% CI 0.74-0.97 per major group). In this cohort disease, excluding hypertension itself, significantly attenuated the age-related rise in systolic blood pressure; the longer disease has been present the less the increase. In addition, socio-economic variables are important predictors of blood pressure change in those with disease. Deprivation index of residential area was a better predictor of disease than previous occupation in these subjects who had retired over a decade previously. Journal of Human Hypertension (2000) 14, 773-778