Staphylococci are major causes of nosocomial blood stream infection. The recently completed SCOPE Surveillance Program found that coagulase-negative staphylococci (CoNS) and Staphylococcus aureus were the first and second most common etiologic agents, respectively, causing nosocomial blood stream infection in the USA. The frequency of oxacillin resistance was 68% among 1553 strains of CoNS and 26% among 787 strains of S. aureus in this study. Extended susceptibility profiles were generated for a subset of 150 S. aureus and 300 CoNS against 16 antimicrobial agents. Oxacillin-susceptible strains of both CoNS and S. aureus were uniformly susceptible to beta-lactam agents with the exception of ampicillin and penicillin. Oxacillin-susceptible S. aureus were also highly susceptible to the fluoroquinolones, aminoglycosides, and trimethoprim/sulfamethoxazole. The oxacillin-susceptible CoNS were less susceptible to these agents, and only glycopeptides were reliably active against oxacillin-resistant strains. PCR detection of the mecA gene was used to scrutinize current NCCLS interpretive breakpoint MICs for determining susceptibility or resistance to oxacillin. We found complete concordance between the presence or absence of mecA and the NCCLS oxacillin interpretive breakpoint categories for S. aureus. In contrast, the NCCLS breakpoints for oxacillin significantly underestimate the degree of true oxacillin resistance among CoNS. Using the presence of mecA as the reference standard, we detected 15.7% false susceptibility to oxacillin using a MIC susceptible breakpoint concentration of < or = 2 micrograms/mL. Lowering the oxacillin MIC breakpoint to < or = 0.25 microgram/mL for CoNS would greatly improve the accuracy of the MIC test performance. We found that both the current oxacillin disk test and the 30-microgram ceftizoxime disk test functioned quite well in predicting those strains of CoNS that contain mecA. These studies have demonstrated both a high level of antimicrobial resistance among nosocomial blood stream isolates of staphylococci as well as significant problems with the current NCCLS breakpoints for oxacillin when testing CoNS.