BACKGROUND: The detection of methicillin resistance is complicated because of its heterogeneous phenotypic expression, and the oxacillin-salt agar screen (OSA) test has been recommended by the National Committee for Clinical Laboratory Standards (NCCLS). The OSA test is recognized as a sensitive and specific test, but the sensitivity and specificity of this test vary among different investigators. In this study, we evaluated the OSA test for detection of methicillin-resistant S. aureus (MRSA) by comparing it to the disk diffusion test and to the MRSA-Screen test (Denka Seiken Co., Ltd., Tokyo, Japan). METHODS: A total of 1,130 S. aureus clinical isolates collected between September 1999 and September 2000 were used. The disk diffusion and OSA tests were carried out on all the isolates according to the NCCLS guidelines. In performing the OSA test, all isolates were inoculated on a growth control plate containing 4% NaCl (without oxacillin). For isolates with discrepant results through the disk diffusion and the OSA tests, the MRSA-Screen test that detects PBP 2a with a monoclonal antibody was performed. RESULTS: Of 1,130 S. aureus isolates, 1,120 isolates (99.1%) yielded concordant results with the disk diffusion and the OSA tests. Ten isolates (0.88%) yielded discrepant results between two methods: three isolates were oxacillin susceptible with the disk diffusion test but grew on the OSA plate. Of these three isolates, two isolates were MRSA-Screen negative. One isolate was oxacillin resistant with the disk diffusion test but did not grow on the OSA plate and was MRSA-Screen positive. Six isolates (0.53%) did not grow either on the OSA plate or the growth control plate. CONCLUSIONS: Only one isolate (0.09%) that was not detectable using the disk diffusion test was correctly classified as MRSA by the OSA test. Also, for accurate detection of MRSA by the OSA test, 4% NaCl-containing growth control plate should be included.