We assessed the intralaboratory imprecision and interlaboratory comparability of lipoprotein measurements and related cardiovascular risk functions such as the total to high density lipoprotein cholesterol (TC:HDL) ratio. Analysis of 5 separate plasma pools was carried out in 4 laboratories which regularly perform lipoprotein testing. We also performed a retrospective audit on RCPA-AACB Quality Assurance Programme data from 134 laboratories participating in the Special Lipid Programme in 1991. Intralaboratory imprecision and interlaboratory comparability are reported as coefficient of variation (cv) and its 95% confidence limit. For the national data, we calculated the percentage of laboratories within specified ranges (±3, 5 or 10%) about the national mean (or median) for a given level of each analyte. Intralaboratory imprecision and interlaboratory comparability amongst the 4 laboratories were close to, or within recommended limits for, TC, TG and HDL, but the interlaboratory comparability of LDL and coronary risk functions exceeded these limits. On a national level, interlaboratory comparability of TC:HDL was within ±5% for only 43% of laboratories, and even fewer (26%) were within this range at higher values of the ratio. We conclude that it is not possible to recommend the use of coronary risk functions at present because of sub-optimal interlaboratory comparability. Even if measures are introduced to overcome this problem, coronary risk functions may over-simplify coronary vascular disease risk in a variety of clinical situations.