Fungal infections are a major problem in patients with hematologic malignancy. Attempts to reduce their frequency with antifungal agents have not been successful. A double-blind, controlled, single-center trial was conducted with 96 consecutive patients undergoing 154 episodes of chemotherapy. Patients received 400 mg of fluconazole or placebo until bone marrow recovery or initiation of intravenous amphotericin B infusions. End points were amphotericin B use, fungal infection, stable neutrophil count > 0.5 x 10(9)/L, toxicity precluding further fluconazole use, and death. By Kaplan-Meier estimation, the time to initiation of amphotericin B therapy was shorter in 76 patients treated with placebo than in 75 treated with fluconazole (P = .003). Also, fluconazole reduced the number of febrile days by 20% (P = .002) and prevented oropharyngeal candidiasis (1/75 vs. 9/76, P = .018). The frequency of deep mycoses (8/76 vs. 8/75) and outcome were unaffected. Fluconazole did not have a favorable effect on infection-related health care costs and was associated with prolonged severe neutropenia (P = .01).