Owing to the lack of a reference technique and of an international cobalamin (vitamin B12) standard, and the large discrepancy between laboratory norms, the authors performed a multicentric study to compare five RIA kits usually used. First, classical tests were used to evaluate the analytical performances of each kit. Results did not demonstrate any superiority of one kit over another. Secondly, B12 values were classified among three categories (low, normal and high) characterized by laboratory and then manufacturer norms. The concordance between these two "judgments" was evaluated with the Kappa coefficient. In addition, the Kappa index proved that the norms supplied by the manufacturer were better than those of laboratories. But mean Kappa coefficient established for each norm gave us an unsatisfactory result. Third, clinical informations allowed to improve the classification of the patients. New limits were defined for each technique and should be tested further, routinely in each laboratory.