OBJECTIVES: Candida species are a common cause of nosocomial bloodstream infection.  Positive blood cultures or high clinical suspicion for candidemia prompt empiric antifungal therapy.  Early species identification allows earlier initiation of appropriate therapy or de-escalation to less expensive narrow spectrum therapy.  Our institute used Candida PNA-FISH (Test A) to rapidly differentiate Candida albicans from non-albicans species. Two newer tests can differentiate among albicans, glabrata or others (Candida PNA-FISH dual probe, Test B) and among albicans/parapsilosis, tropicalis, glabrata/krusei or others (Candida PNA-FISH Traffic Light, Test C). This study aimed to predict cost impacts if newer tests would have been used. METHODS: From a hospital administration perspective, a retrospective chart review of candidemic patients between January 2007 and May 2008 was conducted to obtain costs and utilization of antifungal medications and information on species identification. Monte Carlo simulation models with certain assumptions and limitations were created to predict costs. Model inputs were determined based on real-world data. Only antifungal medication costs, incorporating doses, frequencies and 2008 average wholesale prices, were considered. One-way and probabilistic sensitivity analyses (2,500 trials) were performed. RESULTS: There were 140 candidemic episodes in 132 patients. Candida species isolated included albicans (43%), glabrata (29%), parapsilosis (14%), tropicalis (6%), krusei (6%), and others. Compared to Test A, median potential cost savings per patient are $37 (95% CI $14-$104, Test B) and $51 (95% CI $22-$130, Test C).  Minimal cost savings per patient are $24 (Test B) and $35 (Test C) at a probability of 80%.  Two key variables were identified. Potential cost savings increase with increased empiric use of micafungin or decreased prevalence of Candida albicans. CONCLUSIONS: With sole consideration of antifungal medication costs, switching from Test A to Test C is likely to yield more cost savings than switching to Test B, but cost savings may not be substantial.