The diagnosis and management of adults with hypercholesterolemia in the US are largely based on low-density lipoprotein cholesterol (LDL-C) concentration. In order to classify someone correctly into the National Cholesterol Education Program cut-points, LDL-C must be measured with a total error of =<12%. We examined simultaneously the analytical and clinical performance of two homogeneous LDL-C assays (LDL-C R D , Roche Diagnostics and LDL-C G Z , Genzyme) and the Friedewald calculation (LDL-C F r i e d ). These assays correlated highly with the ultracentrifugation-dextran sulfate-Mg 2 + method (LDL-C R D : r=0.962, y=1.029x-0.48 mmol/l, n=134; LDL-C G Z : r=0.961, y=0.986x-0.12 mmol/l, n=134; LDL-C F r i e d : r=0.960, y=1.017x-0.18 mmol/l, n=115). The total error requirement was met by the LDL-C G Z assay at all clinical decision cut-points, whereas the LDL-C R D assay met this requirement only at LDL-C concentrations of 4.92 mmol/l. The LDL-C F r i e d failed to meet the total error requirement, because the compounded imprecision of the three independent tests required for this calculation was high. Both, the LDL-C G Z and the LDL-C R D assays appeared to be only slightly affected by increasing triglycerides. At the medical decision cut-point range, the LDL-C R D , LDL-C G Z and LDL-C F r i e d assays showed positive predictive values of 89-100, 85-100 and 83-99%, respectively, and negative predictive values of 52-98, 77-98 and 68-98%, respectively. The homogeneous assays provide clinical laboratories with the means to measure LDL-C in hypertriglyceridemic samples and could have a role in the diagnosis and management of hyperlipidemic patients.