A surveillance program (SENTRY) of bloodstream infections (BSI) in the United States, Canada, Latin America, and Europe from 1997 through 1999 detected 1,184 episodes of candidemia in 71 medical centers (32 in the United States, 23 in Europe, 9 in Latin America, and 7 in Canada). Overall, 55% of the yeast BSIs were due to Candida albicans, followed by Candida glabrata and Candida parapsilosis (15%), Candida tropicalis (9%), and miscellaneous Candida spp. (6%). In the United States, 45% of candidemias were due to non-C. albicans species. C. glabrata (21%) was the most common non-C. albicans species in the United States, and the proportion of non-C. albicans BSIs was highest in Latin America (55%). C. albicans accounted for 60% of BSI in Canada and 58% in Europe. C. parapsilosis was the most common non-C. albicans species in Latin America (25%), Canada (16%), and Europe (17%). Isolates of C. albicans, C. parapsilosis, and C. tropicalis were all highly susceptible to fluconazole (97 to 100% at < or =8 microg/ml). Likewise, 97 to 100% of these species were inhibited by < or =1 microg/ml of ravuconazole (concentration at which 50% were inhibited [MIC(50)], 0.007 to 0.03 microg/ml) or voriconazole (MIC(50), 0.007 to 0.06 microg/ml). Both ravuconazole and voriconazole were significantly more active than fluconazole against C. glabrata (MIC(90)s of 0.5 to 1.0 microg/ml versus 16 to 32 microg/ml, respectively). A trend of increased susceptibility of C. glabrata to fluconazole was noted over the three-year period. The percentage of C. glabrata isolates susceptible to fluconazole increased from 48% in 1997 to 84% in 1999, and MIC(50)s decreased from 16 to 4 microg/ml. A similar trend was documented in both the Americas (57 to 84% susceptible) and Europe (22 to 80% susceptible). Some geographic differences in susceptibility to triazole were observed with Canadian isolates generally more susceptible than isolates from the United States and Europe. These observations suggest susceptibility patterns and trends among yeast isolates from BSI and raise additional questions that can be answered only by continued surveillance and clinical investigations of the type reported here (SENTRY Program).