BACKGROUND: Human growth hormone (hGH) is measured for the diagnosis of secretion disorders. These measurements fall under the EU Directive 98/79/EC on in vitro diagnostic medical devices requiring traceability of commercial calibrator values to higher-order reference materials or procedures (Off J Eur Communities 1998 Dec 7;L 331:1?37). External quality assessment schemes show large discrepancies between results from different methods, even though most methods provide results traceable to the recommended International Standard (IS 98/574). The aim of this study was to investigate possible causes for these discrepancies. METHODS: We investigated the commutability and recovery of hGH in reconstituted IS 98/574. We tested different reconstitution protocols and used 4 different serum matrices for spiking. These IS preparations were measured together with serum samples. We quantified hGH by 5 different methods in 4 different laboratories. RESULTS: Results from the different methods correlated well for the serum samples. Mean discrepancies between results from different methods were less than or equal to 20%. None of the IS preparations was commutable for all the method comparisons. The recovery of hGH in preparations of IS 98/574 depended on the reconstitution protocol (>10-fold differences) and background matrix (relative differences less than or equal to17% for different serum matrices). CONCLUSIONS: The use of different protocols for reconstitution and spiking of hGH reference preparations affects quantification by immunoassays, potentially leading to a bias between commercial methods, despite the use of calibrators with values claimed to be traceable to the same higher-order reference material.