Background: Opportunistic invasive fungal infections (IFIs) are a major concern in the management of immunocompromised patients with hematological malignancies. The practice of administering antifungal therapy to neutropenic patients with persistent fever has become a standard care. Conflicting data exists concerning the efficacy of empirical antifungal therapy.Objectives: This study aims to evaluate if empirical antifungal therapy reduces mortality and prevents invasive fungal infections (IFI).Methods: Systematic review and meta-analysis including randomized controlled trials (RCTs) comparing empirical antifungal treatment with placebo or no intervention (control), or another regimen, in neutropenic patients with persistent fever. The Cochrane Library, MEDLINE, conference proceedings and references were searched until 2007. Outcomes assessed were: All-cause mortality, documented IFI, and adverse events. Relative risks (RR) with 95% confidence intervals (CIs) were estimated and pooled.Results: Our search yielded 26 trials, 6 of which compared polyenes or azoles to control, and 20 compared between different regimens of polyenes, azoles or glucan synthesis inhibitors. Compared to control, there was no difference in all-cause mortality (RR 0.98; 95% CI 0.58–1.66, 5 trials, Fig.1). The risk for developing documented IFI was lower (RR 0.23; 95% CI 0.08–0.65, 4 trials, 4 events in the treatment group versus 18 in the control). When azoles (fluconazole, ketoconazole, itraconazole, voriconazole) were compared to polyenes (amphotericin B in 8 trials, liposomal ampho B in 1 trial) there was a trend in favor of azoles for decreased all-cause mortality (RR 0.89; 95% CI 0.73–1.09, 9 trials) and for decreased documented IFI (RR 0.70; 95% CI 0.47–1.04, 8 trials). Adverse events of any kind were less frequent in the azole group (RR 0.40; 95% CI 0.34–0.66, 5 trials), as were those which required discontinuation (RR 0.48; 95% CI 0.38–0.62, 7 trials).Conclusions: Our review demonstrates that empirical antifungal therapy does not reduce mortality. Although it reduces IFIs, data are based on a small number of trials and events. The use of amphotericin B as empirical antifungal therapy seems unwarranted since it appears to be less effective than azoles, with no mortality benefit and an increased rate of side effects. Future trials should pursue a pre-emptive approach using improved diagnostic tools (such as galactomannan testing, high resolution CT), to identify the patients for whom antifungal treatment is warranted. [Display omitted]