Study Objective. To assess the effectiveness, safety, and cost of empiric treatment of febrile neutropenia before and after implementing an algorithm in which voriconazole was substituted for liposomal amphotericin B (L-AmB). Design. Retrospective cohort analysis. Setting. An 850-bed tertiary care hospital, which is also a referral site for patients with acute leukemia. Patients. Fifty-five adult patients who started empiric antifungal therapy for febrile neutropenia between January 1, 2002, and December 31, 2003, encompassing 58 treatment episodes (defined as a hospitalization during which empiric antifungal therapy was administered). Measurements and Main Results. Medical charts, including patients' pharmacy and laboratory data, were reviewed. Twenty-six and 32 episodes of L-AmB and voriconazole use, respectively, were identified. No significant differences between the L-AmB and voriconazole groups were noted at baseline. Rates of fever resolution (54% vs 59%, p=0.791) and breakthrough invasive fungal infections (11% vs 12%, p>0.999) were similar for the L-AmB and voriconazole episodes. Premature drug discontinuation due to the prescriber's perceived lack of efficacy occurred most frequently in the voriconazole group (25% vs 8%, p=0.160). Survival was significantly higher in the voriconazole than in the L-AmB group (100% vs 77%, p=0.006). Adverse effects that were significantly more common in the L-AmB group than in the voriconazole group were elevated serum creatinine levels (27% vs 3%, p=0.017) and electrolyte disturbances (19% vs 0%, p=0.014). Adverse effects reported more frequently in the voriconazole group than in the L-AmB group were visual disturbances (9% vs 0%, p=0.245) and elevated hepatic enzyme levels (9% vs 8%, p>0.999). Mean drug expenditures/episode for initial empiric antifungal therapy were lower for voriconazole than for L-AmB ($1593 vs $4144, or $153 vs $380/day). Conclusion. Our institution's algorithm incorporating voriconazole into the empiric management of febrile neutropenia was associated with effectiveness outcomes comparable to those observed with L-AmB as well as a lower frequency of adverse effects and overall expenditures for antifungal drugs.