The aim of this study was to evaluate the agreement in assay results between commercial kits for the measurement of anti-beta2glycoprotein I antibodies. Ten manufacturers provided one IgG and one IgM kit to three testing centres. Samples from patients with primary (n = 13) or secondary (n = 3) antiphospholipid syndrome (APS), from lupus patients without APS features (n = 6) and from normal individuals (n = 2) were tested in the three centres according to manufacturers' instructions. Dilutions in normal serum of a pool made from positive patients' samples (Forum Calibrators) and dilutions of humanized monoclonal antibodies (MoAbs) were used as additional calibrators. The calibration curves obtained with each calibrator differed widely between kits. The rate of positivity of patients' samples varied from 7 to 16 for IgG and from 2 to 17 for IgM, depending on the kit. Perfect agreement occurred in 12/22 samples for IgG and 5/22 samples for IgM. Samples from normals were found negative by all kits. Between kits, cutoff values varied up to five fold when expressed in Forum Calibrators arbitrary units and up to three fold when expressed in MoAbs equivalents. Examination of discrepant samples indicated that about half of the discrepancies, scoring 8:2 and 9:1, involved the same few kits. In highly discrepant samples, some kits appeared as high responders as compared to others. In conclusion, with the exception of a few kits, agreement in assay results was acceptable. In conclusion, additional efforts are however necessary, especially concerning the way to assess the cutoff point and the adoption of a reference calibrator, in order to improve standardization of the assays.