The relationship between serum total cholesterol, measured at randomization, and mortality was investigated in 822 patients, who were followed for an average of 3.1 years in a double-blind trial, conducted by the European Working Party on High Blood Pressure in the Elderly. Serum cholesterol, measured at randomization, was 0.54 mmol/l higher in women than in men, and declined with increasing age in both men (0.028 mmol/l per year) and women (0.036 mmol/l per year). During follow-up on randomized treatment, cholesterol fell by a similar amount with placebo (0.11 mmol/l per year) and with active treatment (0.14 mmol/l per year). Active treatment consisted of hydrochlorothiazide (25-50 mg/day) plus triamterene (50-100 mg/day) with the addition of alpha-methyldopa (0.5-2.0 g/day) in one-third of the patients. Serum total cholesterol, measured at randomization, was independently and inversely correlated with total (P = 0.03), non-cardiovascular (P = 0.03) and cancer (P = 0.04) mortality during follow-up on double-blind treatment. Total and non-cardiovascular mortality were also negatively correlated with haemoglobin and body weight at randomization.