There is inferential evidence that some patients with prolonged neutropenia and fever not responding to antibacterial agents are at sufficient risk of deep mycoses to warrant empirical therapy, although superiority of an antifungal agent over placebo has not been conclusively demonstrated. Amphotericin B deoxycholate, liposomal amphotericin B, and intravenous itraconazole followed by oral itraconazole solution are licensed in the United States for this indication. Fluconazole and voriconazole have given favorable results in clinical trials of patients with low and high risk of deep mold infections, respectively. Design features that can profoundly influence outcome of empirical trials are (1) inclusion of low-risk patients, (2) failure to blind the study, (3) obscuration of antifungal effects by changing antibacterial antibiotics, (4) failure to balance both arms of the study in terms of patients with prior antifungal prophylaxis or with severe comorbidities, (5) the merging of end points evaluating safety with those of efficacy, and (6) choice of different criteria for resolution of fever.