Detection of oxacillin resistance in coagulase-negative staphylococci (C-NS) by phenotypic methods is often difficult. The present study compared the National Committee for Clinical Laboratory Standards (NCCLS) revised guidelines of phenotypic methods with a mecA-based polymerase chain reaction (PCR) for C-NS. Ninety clinical C-NS isolates were tested for oxacillin resistance by disk diffusion (1-g disk), minimum inhibitory concentration (MIC) breakpoint (0.5g/ml) after 24h, and mecA-based PCR. The sensitivity and specificity of disk diffusion was 80% and 93%, and the sensitivity and specificity of the MIC breakpoint after 24h was 84% and 91%, respectively, against PCR as gold standard. Eleven strains (7 mecA-positive and 4 mecA-negative) showed discordant results between MIC breakpoint after 24h and PCR. Six of the 7 mecA-positive and all 4 mecA-negative discordant strains had inducible oxacillin resistance and -lactamase hyperproduction, respectively. The present study concludes that inducible oxacillin resistance and -lactamase hyperproduction are the major causes of discordant results between phenotypic methods and mecA-based PCR, and need special attention.