The complete lipoprotein profile is thought to give more information about the individual risk of coronary heart disease than total cholesterol alone. Although total cholesterol has a low sensitivity in the correct assessment of the risk of coronary heart disease, it may be of value in screening programs because of its low cost. In this study of 5,335 subjects, total cholesterol gave a different assessment of coronary heart disease risk (United States National Cholesterol Education Program guidelines) in 25% of subjects than the complete lipoprotein profile. Differences in risk assignment were mainly accounted for by high- and low-density lipoprotein-cholesterol (Friedewald equation). The calculated low-density lipoprotein-cholesterol was highly correlated with the value measured with a mixed ultracentrifugation and precipitation procedure. However, calculated values gave estimates of coronary heart disease risk which were 20% different from those from measure values. In 200 subjects in whom the lipoprotein profile was assessed three times in 1 year, the total cholesterol low-density lipoprotein-cholesterol varied by more than 30 mg/dl (0.78 mmol/l) in 52% and 50%, respectively, triglycerides by more than 30 mg/dl (0.34 mmol/l) in 75%, and high-density lipoprotein-cholesterol by more than 15 mg/dl (0.39 mmol/l) in 34%. Compared with the mean of the measurements, the single measurement of total cholesterol misclassified 48% of subjects, low-density lipoprotein-cholesterol 60%, high-density lipoprotein-cholesterol 12%, and 28%. We conclude that total cholesterol alone may be misleading in the assignment of coronary heart disease risk. Calculation of low-density lipoprotein-cholesterol, although less accurate than desirable, is the only way of evaluating this in clinical practice.(ABSTRACT TRUNCATED AT 250 WORDS)