Follow-up studies of men in the United States reveal that the risk of coronary attack is not the same for all and that the predisease cholesterol level is a useful and quantitative predictor of relative risk. An insurance medical approach to this information is developed in which a crude classification of mortality risk is made according to serum cholesterol level. This is devised as a temporary aid to those interested in risk appraisal of men in apparent good health and will be replaced within a few years by quantitative data obtained from follow-up studies of insured persons in the United States and Britain. At present, serum cholesterol level, as an isolated factor, is not widely used as a risk variable by insurance underwriters. On the basis of average figures from published follow-up studies, and with a number of assumptions, a practical "standard" risk category is assigned to men in the United States with serum cholesterol values between 200 and 274 mg.%, "better than standard" risk below 200 mg.%, and "excess" risk rising exponentially from 50 per cent extra (150% of standard) between 275 and 300 mg.%, to twice standard at 300 to 349 mg.%, triple standard at 350 to 399 mg.%, and four times standard risk (400%) at cholesterol levels of 400 mg.% and beyond. The effect of variability of cholesterol values on misclassification is considered, and it is recommended that an average value be obtained from several determinations when serum cholesterol is to be used as a factor in risk classification. Evidence is reviewed which indicates that at this stage of knowledge there is no single technic superior to measurement of blood cholesterol during health for the prediction of subsequent risk of coronary attack, and there is none more reproducible, available, or adaptable to the purposes of both insurance and preventive medicine.