The objective of this trial was to evaluate the potential advantages of the combination of piperacillin and tazobactam in the control of fever in neutropenic patients. In this single-center study, patients who experienced a total of 247 febrile episodes were prospectively randomized to receive either our standard regimen, ceftazidime 3 g/day (1 g t.i.d.) plus tobramycin 3 mg/kg per day (1.5 mg/kg b.i.d.), or piperacillin 12 g/day plus tazobactam 1.5 g/day (4 g+0.5 g t.i.d.) plus tobramycin 3 mg/kg per day (1.5 mg/kg b.i.d.). Vancomycin was added in all cases of persistent fever in the ceftazidime arm, but only when there was microbiologically documented resistance in the piperacillin/tazobactam arm. All 247 episodes were evaluable by "intent-to-treat" analysis. The two populations were well matched in terms of age, gender, underlying disease, chemotherapy received, oral decontamination, clinical and bacterial documentation, and severity and duration of neutropenia. Initial antibacterial therapy was successful (apyrexia at 72 h, without antibiotic change) more frequently (P = 0.008) with the regimen containing piperacillin/tazobactam (54.4%) than with the one including ceftazidime (37.6%). Fewer (P = 0.02) major infectious events (infectious death or delay in treatment of underlying disease due to infection) were observed during piperacillin/ tazobactam treatment (2.6%) than with the ceftazidime regimen (11.3%), despite a lower frequency of glycopeptide addition when piperacillin/tazobactam was used (54.4% versus 77.4%) according to the rules adopted. This trial confirmed the efficacy of the piperacillin/tazobactam combination for empirical treatment of febrile neutropenic patients. This antibiotic combination permitted a dramatic decrease in empiric glycopeptide antibiotic administration in such patients.