Candida bloodstream infection (CBSI) accounted for 50% of bloodstream infections in our medical intensive care unit (MICU) in 2004. Our objective was to evaluate a risk-based fluconazole prophylaxis program. CBSI incidence, patient demographics, and unit metrics were retrospectively reviewed for 2004. Starting on January 2005, patients meeting pre-specified criteria were placed on risk-based fluconazole prophylaxis and their outcomes, adverse events, and unit metrics were prospectively collected. The inclusion criteria were based on a clinical prediction rule and included an MICU stay greater than 72 h, broad-spectrum antibiotics, and central venous catheter, along with at least two of the following: mechanical ventilation for at least 48 h, any type of dialysis, parenteral nutrition, pancreatitis, systemic steroids, or other systemic immunosuppressive agents. For 2004, the unit had nine CBSI, corresponding to a rate of 3.4 CBSI/1,000 line-days. Four cases were caused by C. albicans, four by C. glabrata, and one by C. tropicalis. The mean +/- standard deviation (SD) APACHE II score for these patients was 25 +/- 9. In 2005, a total of 36 patients (2.6% of all unit admissions) received prophylaxis and the unit had two CBSI, corresponding to a rate of 0.79 CBSI/1,000 line-days. One patient had C. albicans and the other had C. tropicalis. The mean +/- SD APACHE II score for these patients was 21 +/- 8. The mean +/- SD duration of fluconazole prophylaxis was 8 +/- 6 days. Fluconazole was discontinued in two patients due to non-severe adverse events (acute eosinophilia, elevated transaminases). The attributable cost of CBSI in the unit in 2004 was $63,000 per episode. The total cost for the 36 courses of fluconazole was $6,000. When comparing the 2004 CBSI patients and the 2005 prophylaxis patients, we found similar acuity, demographics, and risk factors, with no differences in MICU or hospital mortality or length of stay. Risk-based fluconazole prophylaxis in an MICU with a high incidence of CBSI was safe and cost-effective when applied to a limited number of patients and produced a significant decrease in the incidence of this disease.