Lipoprotein cholesterol concentrations in plasma are routinely estimated by using the Friedewald formula, whereby very-low-density lipoprotein cholesterol (VLDL-C) is estimated to be one-fifth the plasma triglyceride concentration. Ordinarily, this formula is applied only to plasma sampled from patients in the fasted state. To determine whether lipoprotein cholesterol measurements are altered substantially in plasma sampled from nonfasting subjects, we obtained postprandial blood samples from 22 healthy subjects (nine men, 13 women, ages 22-79 years) fed a fat-rich meal (1 g fat per kilogram body wt.). The plasma triglyceride concentration increased postprandially in all subjects (233 +/- 16% of baseline at 3 h). The mean cholesterol concentration in plasma was essentially unchanged. High-density lipoprotein cholesterol (HDL-C) was significantly decreased (94 +/- 2% at 3 h, P less than 0.001). VLDL-C and low-density lipoprotein cholesterol (LDL-C), estimated by the Friedewald formula, were compared with measurements obtained by modified Lipid Research Clinics (LRC) methodology. As measured by either method, VLDL-C increased and LDL-C decreased significantly after the fat-rich meal. These postprandial changes were significantly greater (P less than 0.01) when estimated by the Friedewald formula than by LRC methodology. We conclude that (a) lipoprotein cholesterol concentrations measured in the fed subject differ significantly from those measured in the fasted subject, and (b) plasma must be obtained after at least a 12-h fast if an individual's risk of coronary heart disease is to be accurately assessed.