In a randomized study, caspofungin was compared with amphotericin B for the treatment of invasive candidiasis in a total of 239 adults from 56 sites in 20 countries. This study provided a unique opportunity to assess the frequency and outcome of invasive candidiasis caused by different Candida species worldwide, and the results are presented here. Efficacy was primarily assessed at the end of intravenous therapy using a modified intent-to-treat (MITT) analysis. This analysis was performed on 224 of the 239 patients enrolled in the study. Attempts were made to collect baseline Candida isolates from all patients for species identification at a central laboratory. Yeasts were identified to the species level using two commercial systems and microscopic examination. Viable baseline isolates were recovered from 210 of the 224 (94%) patients included in the MITT analysis. Candida albicans was the most frequently isolated species in all regions and was responsible for 45% of cases overall. Nevertheless, the majority of cases of infection were caused by non- albicans Candida species. In the USA and Canada, Candida glabrata was the second most commonly isolated pathogen (18%). In contrast, Candida parapsilosis and Candida tropicalis accounted for 55% of cases in Latin America. Outcomes were comparable for patients treated with caspofungin (74% overall; 64% and 80% for infections due to Candida albicans and non- albicans species) and amphotericin B (62% overall; 58% and 68% for infections due to Candida albicans and non- albicans species), and were generally similar across continents. The distribution of Candida species isolated from patients enrolled in a clinical trial may not be representative of pathogens causing invasive candidiasis in the general population. Nevertheless, our findings may affect the regional choice of empirical antifungal therapy for seriously ill patients with suspected or documented invasive candidiasis since different Candida species have varying susceptibility to conventional antifungal drugs.