The risks associated with various levels of serum cholesterol were determined by analysis of 6-year mortality in 361 662 men aged 35-57. Above the 20th percentile for serum cholesterol (> 181 mg/dl, >4·68 mmol/l), coronary heart disease (CHD) mortality increased progressively; the relative risk was large (3·8) in the men with cholesterol levels above the 85th percentile (>253 mg/dl, > 6·54 mmol/l). When men below the 20th cholesterol percentile were used as the baseline risk group, half of all CHD deaths were associated with raised serum cholesterol concentrations; half of these excess deaths were in men with cholesterol levels above the 85th percentile. For both CHD and total mortality, serum cholesterol was similar to diastolic blood pressure in the shape of the risk curve and in the size of the high-risk group. This new evidence supports the policy of a moderate fat intake for the general population and intensive treatment for those at high risk. There is a striking analogy between serum cholesterol and blood pressure in the epidemiological basis for identifying a large segment of the population (10-15%) for intensive treatment.