Broth microdilution was compared with tube macrodilution and a simplified population analysis agar method for evaluating vancomycin and teicoplanin MICs and detecting glycopeptide-intermediate isolates of Staphylococcus aureus. Modal vancomycin and teicoplanin MICs recorded by tube macrodilution and the agar plate assay, which both used inocula of 10(6) CFU, were significantly higher (2 microg/ml) against a panel of borderline glycopeptide-susceptible and glycopeptide-intermediate methicillin-resistant S. aureus (MRSA) bloodstream isolates compared to broth microdilution (1 microg/ml). Vancomycin and teicoplanin MIC distributions by tube macrodilution and agar testing were also markedly different from those evaluated by broth microdilution. The 20-fold-lower inoculum size used for broth microdilution compared to macrodilution and agar MIC assays explained in part, but not entirely, the systematic trend toward lower vancomycin and teicoplanin MICs by microdilution compared to other methods. Broth microdilution assay led to underdetection of the vancomycin-intermediate S. aureus (VISA) phenotype, yielding only three VISA isolates, for which vancomycin MICs were 4 microg/ml compared to 8 and 19 VISA isolates detected by macrodilution and agar testing, respectively. While macrodilution and agar testing detected 7 and 22 isolates with elevated teicoplanin MICs (8 microg/ml), respectively, broth microdilution failed to detect such isolates. Detection rates of isolates with elevated vancomycin and teicoplanin MICs by macrodilution and agar testing assays were higher at 48 h than at 24 h. In conclusion, the sensitivity of broth microdilution MIC testing is questionable for reliable detection and epidemiological surveys of glycopeptide-intermediate resistance in S. aureus isolates.