According to the new guidelines of the National Cholesterol Education Program (NCEP) for secondary prevention in adults with evidence of coronary heart disease or other clinical atherosclerotic disease, lipoprotein analysis is required and classification is based on low density lipoprotein (LDL) cholesterol. The aim of the present study was to analyze the reliability of calculated LDL cholesterol by the Friedewald formula compared with measured LDL cholesterol after separation by ultracentrifugation in 202 male patients with extracoronary atherosclerosis (100 patients with ischemic cerebrovascular disease and 102 patients with peripheral vascular disease) and in 117 healthy control subjects. Calculated LDL cholesterol coincided with measured LDL cholesterol, with less than 10% error, in 118 patients (58.4%) with extracoronary atherosclerosis and in 87 controls (74.4%). Calculated LDL cholesterol was overestimated, with an error of 10% or more compared with measured LDL cholesterol, in 34.6% of patients and 22.2% of controls, and underestimated in 6.9% and 3.4%, respectively. Despite a good correlation between calculated and measured LDL cholesterol, the intraclass correlation coefficients demonstrate a poor concordance between calculated and measured LDL cholesterol, both in patients and controls. The authors underline the need for caution in assessing the reliability of calculated LDL cholesterol.