A selected method for the determination of creatinine in plasma, using the reaction with alkaline picrate without prior pretreatment has been proposed by the Commission 'Validation de techniques' in the SFBC (Société Française de biologie clinique). The transferability step was conducted in seven laboratories, equipped with different automatic analyzers, using analytical procedures derived from the recommended method. Its goal was to test whether the original analytical performances could be maintained and consistent results obtained. The validation step was designed to evaluate the linearity limits of the analytical range, the detection limit, to assess accuracy as compared to a high performance liquid chromatography and to investigate the effect of the main interferents. Linearity limits are 15 and 2000 mumol/L. The detection limit is 3 to 8 mumol/L according to the analytical systems. The selected method can fulfil the set imprecision goals: intralaboratory CV minus than 2% (within-run), minus than 4% (run-to-run), interlaboratory CV minus than 5% (for 100 mumol/L creatinine). Inaccuracy evaluated for the chosen control sera is 1 to 15% as compared to the chromatographic method, according to the sera and to the analytical systems. The results obtained with the selected method are more consistent with the HPLC than are those obtained with an alkaline picrate method without SDS or with an enzymatic method. No interference could be demonstrated for acetoacetate (up to 8 mmol/L), hemoglobin (up to 210 mumol/L), unconjugated bilirubin (up to 250 mumol/L), glucose (up to 30 mmol/L), IgG (up to 45 g/L), albumin (up to 60 g/L). The effect of cephalosporins depends on the molecule. The reagents are stable for at least 6 months when stored in closed vials at +20 degrees C. The alkaline reagent is stable 30 days at +4 degrees C. Reference limits (0.025 and 0.975 fractiles) have been established for healthy adults. They are respectively 73 to 126 mumol/L for men and 59 to 100 mumol/L for females.