Using data from over 10 000 men, women, and children who participated in the Lipid Research Clinics prevalence studies, we have examined the formula adopted by Friedewald et al1 for estimating plasma or serum concentrations of low-density lipoprotein cholesterol (LDL-C) when (for economy, or in the absence of an ultracentrifuge) only fasting total cholesterol (TC), high-density lipoprotein cholesterol (HDL-C), and triglyceride (TG) concentrations are measured in milligrams per liter, ie, LDL-C=TC—(HDL-C+0.20XTG). Values for LDL-C obtained by use of the Friedewald formula were compared with values derived from the Lipid Research Clinics ultracentrifugal procedure for LDL-C, which was used as a reference. Participants who were pregnant, who had not fasted, or whose plasma contained chylomicrons or floating 0-lipoproteins were excluded. We concluded that a better estimator for LDL-C was provided by the equation LDL-C-TC—(HDL-C+0.16XTG), since it produced an error (relative to the reference method) of lesser magnitude than the previous formula. The expression 0.16XTG (0.37XTG when measurements are reported in millimoles per liter) also produced a more accurate estimate of very low-density lipoprotein cholesterol relative to values obtained by the standard Lipid Research Clinics procedure for this component. The proposed formula is more precise for plasmas or sera with a TG concentration within the normal range.