Systemic fungal infection is considered to be an important cause of morbidity and mortality in cancer patients, particularly those with neutropenia. Antifungal drugs are often given prophylactically, or to patients with persistent fever. The objective of this review was to assess the effect of antifungal drugs in cancer patients with neutropenia. We searched the Cochrane Controlled Trials Register and MEDLINE (November 2001) and the reference lists of articles. We searched the proceedings of the ICAAC (from 1990 to 2001, General Meeting of the ASM (from 1990 to 2001), and the European Congress of Clinical Microbiology and Infectious Diseases (1995 to 2001) and contacted researchers in the field. Randomised trials of amphotericin B, fluconazole, ketoconazole, miconazole, or itraconazole compared with placebo or no treatment in cancer patients with neutropenia. Two reviewers independently assessed trial eligibility, methodological quality and abstracted data. Thirty trials involving 4094 patients were included. Prophylactic or empirical treatment with antifungals as a group had no statistically significant effect on mortality (relative risk 0.95, 95% confidence interval 0.82 to 1.11). The relative risk was smallest for amphotericin B, 0.73 (0.52 to 1.03) (P=0.08). In another review, three trials compared intravenous lipid soluble amphotericin B (AmBisome) with smaller doses of standard intravenous amphotericin B; the relative risk was 0.74 (0.52 to 1.07). Taken together, these results indicate that intravenous amphotericin B might decrease mortality. In contrast, trials with fluconazole, ketoconazole, miconazole and itraconazole failed to find an effect on mortality. The incidence of invasive fungal infection decreased significantly with administration of amphotericin B (relative risk 0.39, 95% CI 0.20 to 0.76), fluconazole (0.39, 0.27 to 0.57) and itraconazole (0.51, 0.27 to 0.96), but not with miconazole or ketoconazole. Intravenous amphotericin B is the only antifungal agent for which there is evidence suggesting that it might reduce mortality. It should therefore be preferred when prophylactic or empirical antifungal therapy in cancer patients with neutropenia is considered indicated.