The performance of approximately 400 participants in the CAP Special Mycology Survey in identifying 15 medically important yeasts was analyzed. With nine strains, users of the API 20C system and Uni-Yeast-Tek (UYT) systems performed comparably. With two strains (Trichosporon cutaneum and Candida guilliermondii) the UYT users performed significantly better than API users; with four strains (Cryptococcus albidus, Saccharomyces cerevisiae, and two strains of Candida tropicalis), API users significantly outperformed UYT users. Analysis of biochemical reaction patterns demonstrated, with a few exceptions, good reproducibility among laboratories with both kits. With one exception (melezitose-variable Candida albicans), the few discrepancies in biochemical reactions had little or no influence on the final identification. From this analysis, we conclude that both API and UYT perform acceptably well in routine clinical laboratories when used in conjunction with morphologic observations and, when indicated, supplemented biochemical test.