Seventeen laboratories in the British Isles participated in a study to compare six different commercially available immunoassays for serum lipoprotein(a) (Lp(a)) and to establish reasons for the variations in the measurement of serum Lp(a) concentrations. Pooled serum was distributed neat and after dilution at a central laboratory. In addition, the central laboratory sent unpooled serum sampled monthly from six healthy volunteers to each of the participating laboratories for 12 months. The assays all gave linear dilution curves which were parallel, although the reported values varied twofold. There were major differences in the values assigned to different manufacturers' calibrants which was not explained by whether the units employed were whole Lp(a), the protein moiety of Lp(a) or simply apolipoprotein(a). The coefficient of variation for the reported value of Lp(a) over 12 months was 33%. The component variation was 10% after adjustment for inter-laboratory and intra-laboratory variation. Some individuals clearly had a greater tendency to variable serum Lp(a) concentrations than others, but all the assays responded to this in the same way. Thus, the assays tested probably measured the same analyte. The problem of calibration could largely be addressed if agreement were reached by the manufacturers. Even with improvements in analytical precision it should be realized that multiple measurements of serum Lp(a) levels are necessary if the true mean value is to be appreciated. Individuals showing wide variation in serum Lp(a) may reward further study if its role is to be established.