Purpose: To evaluate three methods for 406 isolates of Staphylococcus aureus and coagulase-negative staphylococci (CNS) for the detection of methicillin resistance (MR) using National Committee for Clinical Laboratory Standards (NCCLS) new interpretive criteria. Methods: We used polymerase chain reaction (PCR) as a gold standard method to evaluate three methods [disk diffusion with Mueller-Hinton agar (MHA) and mannitol salt agar (MSA) and Sceptor system (Becton Dickinson, USA)] for the detection of mecA gene. The isolates that were methicillin-resistant with any of the three tests were evaluated further for MR by E-test. Results: MHA, MSA and Sceptor showed sensitivities of 100, 100 and 99 for S. aureus and 100, 82.6 and 72.1% for CNS, respectively. The specificities of the same methods were found as 100, 90.1 and 99.3% for S. aureus and 79.2, 95.8 and 97.2% for CNS, respectively. E-test showed 100 sensitivity for both S. aureus and CNS. Forty-eight CNS and 16 S. aureus isolates, which presented discrepancies with the three phenotypic methods (MHA disk diffusion method, MSA disk diffusion method and Sceptor), were correctly classified as resistant/susceptible with the E-test when compared with PCR. Only five CNS isolates, which were mecA-negative with PCR were resistant with E-test. Analysis of 248 S. aureus revealed that MHA is superior to other phenotype-based susceptibility testing methods in detecting MR. When we examined the results of 158 CNS, none of the three methods proved efficient in detecting MR. Conclusions: We conclude that although the accuracy of the MHA disk diffusion test for the detection of MR approaches the accuracy of PCR for S. aureus isolates, the need for easy and reliable methods of detecting MR in CNS still remains.