BACKGROUND: Patients with neutropenia and persistent fever are often treated empirically with amphotericin B or liposomal amphotericin B to prevent invasive fungal infections. Antifungal triazoles offer a potentially safer and effective alternative. METHODS: In a randomized, international, multicenter trial, we compared voriconazole, a new second-generation triazole, with liposomal amphotericin B for empirical antifungal therapy. RESULTS: A total of 837 patients (415 assigned to voriconazole and 422 to liposomal amphotericin B) were evaluated for success of treatment. The overall success rates were 26.0 percent with voriconazole and 30.6 percent with liposomal amphotericin B (95 percent confidence interval for the difference, -10.6 to 1.6 percentage points); these rates were independent of the administration of antifungal prophylaxis or the use of colony-stimulating factors. There were fewer documented breakthrough fungal infections in patients treated with voriconazole than in those treated with liposomal amphotericin B (8 [1.9 percent] vs. 21 [5.0 percent], P=0.02). The voriconazole group had fewer cases of severe infusion-related reactions (P<0.01) and of nephrotoxicity (P<0.001). The incidence of hepatotoxicity was similar in the two groups. Patients receiving voriconazole had more episodes of transient visual changes than those receiving liposomal amphotericin B (22 percent vs. 1 percent, P<0.001) and more hallucinations (4.3 percent vs. 0.5 percent, P<0.001). Parenteral voriconazole was changed to the oral formulation in 22 percent of the voriconazole group, with a reduction in the mean duration of hospitalization by one day in all patients (P=0.17) but by two days in patients at high risk (P=0.03). CONCLUSIONS: Voriconazole is a suitable alternative to amphotericin B preparations for empirical antifungal therapy in patients with neutropenia and persistent fever.