An improved fluconazole 25-mg disk diffusion method was used to test the susceptibility of 20,900 consecutively isolated clinical strains of Candida species from 40 hospital laboratories in 26 countries. The procedure is similar to the National Committee for Clinical Laboratory Standards (NCCLS) M2-A6 method for testing bacteria, except Mueller-Hinton agar is supplemented with 2% glucose and 0.5 mcg/mL methylene blue. Plates were incubated at 35 degrees C and read after 18 to 24h. Tentative zone interpretive criteria were based on the correlation by regression analysis with the NCCLS M27-A Reference Method for Broth Dilution Antifungal Susceptibility Testing of Yeasts: > =19mm Susceptible, < = 12mm Resistant, and 13-18 mm Susceptible-Dose Dependent. Of 14,368 isolates of Candida albicans, 2,073 C. glabrata, 869 C. tropicalis, 752 C. parapsilosis, and 351 C. krusei, 99%, 67%, 90%, 94%, and 26%, respectively, were susceptible in vitro to fluconazole. All study sites used the BIOMIC System to electronically read zones on plates, interpret, record test results and verify quality control data. This is the largest study to date that evaluated a broad range of sequentially collected yeasts from various infections and different types of hospitals. The modified disk testing procedure is facile and economical to perform and offers a reproducible and accurate means to assess the in vitro susceptibility of Candida species to fluconazole.