The relation between serum cholesterol and mortality was investigated in 822 elderly hypertensive patients randomly assigned to treatment with diuretics, with or without methyldopa, or placebo. Cox's proportional hazards model showed that pretreatment serum total cholesterol levels were independently and inversely correlated with total mortality (p = 0.03), noncardiovascular mortality (p = 0.02), and cancer mortality (p = 0.04) during treatment. Total and noncardiovascular mortality were also negatively correlated with pretreatment hemoglobin levels and body weight. All factors being equal, an increase in total serum cholesterol of 2.3 mmol/L was associated with a one-year prolongation of survival. After adjustment for gender, age, pretreatment cardiovascular complications, and systolic pressure, the correlations between serum cholesterol and cardiovascular and cardiac mortality were not significant.