Résumé : The reproducibility of the binary typing (BT) protocol developed for epidemiological typing of Staphylococcus aureus was analyzed in a biphasic multicenter study. In a Dutch multicenter pilot study, 10 genetically unique isolates of methicillin-resistant S. aureus (MRSA) were characterized by the BT assay as presented by van Leeuwen et al. [J. Clin. Microbiol. 2001 39 (1) 328]. The BT assay, including a standardized DNA extraction protocol was performed in duplicate in eleven medical microbiology laboratories. Two different hybridization detection procedures were applied and a prelabeled DNA sample as process control was included. Only three laboratories accurately identified all strains. Divergence in technical procedures resulted in misinterpretation due to an increasing number of faint or absent hybridization signals in combination with high background staining. The binary type of the process control was determined correctly by all participating laboratories. The feasibility of the BT protocol was related directly to the skill of the laboratory personnel. On the basis of the national study, we concluded that the DNA extraction protocol needed modification to improve DNA yield and purity. Subsequently, seven European laboratories participated in an international study to determine the reproducibility of the modified BT protocol. Each center was asked to analyze 10 DNA samples previously extracted from 10 MRSA strains (phase 1) and, additionally, to analyze 10 MRSA strains, using the standardized or their in-house DNA isolation protocol (phase 2). A prelabeled DNA process control sample was included again. The binary types of all DNA samples were identified correctly by all but one laboratories. This latter laboratory diverged from the protocol by adding an excess of labeled DNA to the hybridization mixture, resulting in a high background and, therefore, noninterpretable BT results. All centers produced identical BT results for the process control. Five of the seven centers correctly identified the binary types of all 10 MRSA strains in phase 2 of the international study. Three of these centers used their in-house DNA extraction protocol. Divergence from the standard BT protocol in the remaining two centers resulted in no interpretable BT data for the 10 MRSA strains. The study demonstrated that each center that followed the BT protocol to the letter could generate reproducible results, irrespective whether or not an in-house DNA isolation protocol was used. The current BT protocol thus represents a simple method generating robust, reproducible genotype data for S. aureus strains. © 2002 Elsevier Science B.V. All rights reserved.