In a study with young children receiving growth hormone treatment, it was necessary to re-establish reference values for various blood components, e.g. apolipoproteins A-I and B. We find that considerable method-to-method variation still exists. We used the DuPont Dimension analyzer (D) and the Beckman Array analyzer (B) systems and a third procedure with Orion reagents and a Kone Analyzer (K). We investigated if assay results may be pooled or exchanged within our study. In 59 serum samples we measured apolipoprotein A-I and apolipoprotein B (n = 58) and calculated the orthogonal regression equations y = a(Sa) x + b(Sb). For apolipoprotein A-I the results are: (I) B = 1.165 (0.065) D - 0.193 (0.077), Syx = 0.055; with r = 0.922 and (II) K = 0.831 (0.056) D - 0.190 (0.066), Syx = 0.055; with r = 0.898. For apolipoprotein B the equations are: (III) B = 1.586 (0.137) D - 0.246 (0.100), Syx = 0.061; with r = 0.840 and (IV) K = 0.869 (0.065) D + 0.251 (0.048), Syx = 0.044; with r = 0.875. According to Passing & Bablok, the slopes and intercept values are 1.093 and -0.126; 0.848 and 0.167; 1.500 and -0.185; 0.880 and 0.249. The overall impression is the same for both regression methods: comparability had not yet been achieved by early 1993, particularly not for apolipoprotein B, and reference values differ significantly depending on the selected methodology.