During a multicenter evaluation, 16 methods for creatinine measurement have been tested according to the guidelines of the Société française de biologie clinique (SFBC) protocol. Kinetic Jaffé methods, widely used in France, performed on different analytical systems (Astra Beckman, IL 508, RA 1000 Technicon, Hitachi 704, 705, 717 Boehringer, Fara Roche, Progress Kone, Kem-O-Mat Coulter, Perspective France Monitor) have been compared to a continuous flow method with aqueous standards, to enzymatic methods using creatinine amidohydrolase with a colorimetric measurement (Boehringer and Ektachem Kodak) and to an HPLC method. Reproducibility, estimated with four different control sera, proved to be unsatisfactory in some cases as compared to current criteria for imprecision (less than +/- 10 mumol/l for intralaboratory and less than +/- 20 mumol/l for interlaboratory imprecision). The same selected patients sera covering the whole range of physiopathological concentrations have been analyzed with each method, and compared with the continuous flow results. Differences are more dependent on the sample than on the calibrators. The influences of haemolysis, bilirubin, acetoacetate, albumin, lipids, glucose, and some cephalosporins have been evaluated with spiked human sera. Haemolysed, turbid and jaundiced patient samples have been analyzed as well. The results vary according to the analytical procedure. This study took place in the implementation of a selected method for routine purpose with special regards to interferences and an acceptable imprecision. The method must satisfy the physicians' demands in the renal function exploration, especially in kidney-transplant patients.