A multilaboratory study was undertaken to determine the accuracy of the current National Committee for Clinical Laboratory Standards (NCCLS) oxacillin breakpoints for broth microdilution and disk diffusion testing of coagulase-negative staphylococci (CoNS) by using a PCR assay for mecA as the reference method. Fifty well-characterized strains of CoNS were tested for oxacillin susceptibility by the NCCLS broth microdilution and disk diffusion procedures in 11 laboratories. In addition, organisms were inoculated onto a pair of commercially prepared oxacillin agar screen plates containing 6 microg of oxacillin per ml and 4% NaCl. The results of this study and of several other published reports suggest that, in order to reliably detect the presence of resistance mediated by mecA, the oxacillin MIC breakpoint for defining resistance in CoNS should be lowered from >/=4 to >/=0.5 microg/ml and the breakpoint for susceptibility should be lowered from </=2 to </=0.25 microg/ml. In addition, a single disk diffusion breakpoint of </=17 mm for resistance and >/=18 mm for susceptibility is suggested. Due to the poor sensitivity of the oxacillin agar screen plate for predicting resistance in this study, this test can no longer be recommended for use with CoNS. The proposed interpretive criteria for testing CoNS have been adopted by the NCCLS.