We evaluated the performance of Etest using several different agar media for testing of amphotericin B against 660 clinical isolates of yeast species including Candida albicans, C. glabrata, C. tropicalis, C. parapsilosis, C. lusitaniae, C. krusei, Candida spp., Cryptococcus neoformans, and Saccharomyces cerevisiae. Two of the C. lusitaniae isolates represented strains with high-level amphotericin B resistance. All isolates were tested by NCCLS microdilution methods with RPMI 1640 medium and by Etest using RPMI agar with 2% glucose (RPG). A subset of 108 isolates was also tested by Etest using RPG, antibiotic medium 3 agar (AM3), Casitone agar (CAS), and Mueller-Hinton agar (MHA). The overall agreement between the NCCLS reference method and Etest using RPG was 98.3%. All of the Etest methods identified the two resistant strains (MICs, 4.0 to 16 micrograms/mL), whereas the reference method failed to distinguish them from 18 other isolates with MICs of 2.0 micrograms/mL. Among the 20 isolates with reference MICs of 2.0 micrograms/mL, 12 had MICs > or = 2.0 micrograms/mL when tested by Etest with RPG (range 2.0 to 16 micrograms/mL) compared with eight with AM3, two with CAS, and five with MHA. These data indicate that Etest identifies subpopulations of yeast isolates with high amphotericin B MICs. The greater sensitivity of Etest for detection of amphotericin B resistance should be exploited in future surveillance studies.