An evaluation was undertaken to determine the optimal method for testing the susceptibilities of 100 clinical isolates and two reference strains of Enterococcus spp. to vancomycin in vitro. Six testing methods were studied by using the following media and incubation times: agar screen with the Synergy Quad Plate (Remel, Lenexa, Kans.), an in-house-prepared brain heart infusion (BHI) agar plate, and an in-house-prepared Mueller-Hinton (MH) agar plate, all incubated for 24 or 48 h; broth microdilution (Sensititre Just One Strip; AccuMed International, Inc., West Lake, Ohio) with BHI or cation-adjusted MH broth incubated for 24 or 48 h; agar dilution with BHI or MH agar incubated for 24 or 48 h; epsilometer test (E test; AB BioDisk, Solna, Sweden) with BHI or MH agar incubated for 24 or 48 h; disk diffusion with BHI or MH agar incubated for 24 or 48 h; and the automated Vitek method with the gram-positive susceptibility Staphylococcus aureus card and R02.03 software (bioMerieux, Inc., Hazelwood, Mo.). Growth failures occurred with MH media (n = 6) but not with BHI media. One growth failure occurred with the Vitek method. Results for each testing method for each Enterococcus strain were interpreted as susceptible, intermediate, or resistant according to current National Committee for Clinical Laboratory Standards (NCCLS) criteria and compared to the vancomycin resistance genotype (i.e., vanA, vanB, vanC-1, or vanC-2/3). For all methods, extension of the incubation time from 24 h to 48 h either produced no difference in the results or gave poorer results. The following methods produced no very major or major interpretive errors: broth microdilution with BHI media incubated for 24 h, agar dilution with BHI media incubated for 24 or 48 h, and E test with BHI media incubated for 24 or 48 h. Unacceptable frequencies of very major errors (> 1%) occurred with all methods for which MH media were used. Minor interpretive errors were frequent with all methods. These minor interpretive errors also occurred most frequently with Enterococcus strains with vanC genes, which encoded low-level vancomycin resistance (MIC < or = 8 microg/ml), as opposed to Enterococcus strains which possessed vanA or vanB genes, which encoded higher-level vancomycin resistance (MIC > or = 64 microg/ml). Modification of NCCLS breakpoints, especially for motile Enterococcus spp. (E. casseliflavus, E. flavescens, and E. gallinarum), may resolve this problem; however, in the current study, one E. faecalis strain and one E. faecium strain carried only the vanC gene. The agar screen method may also require reformulation. The current agar screen plate contains 6 microg of vancomycin per ml, which may not detect all low-level resistance associated with vanC genotypes. Nevertheless, the clinical significance of this low-level vancomycin resistance remains unknown.