Between July 1993 and September 1996, 107 consecutive febrile episodes in 83 neutropenic cancer patients with a median age of 41 years were randomized to treatment either with piperacillin/tazobactam 4.5 g every 8 h i.v. or ceftazidime 2 g every 8 h plus amikacin 15 mg/kg i.v. per day. In the case of fever >38° C 48 h after initiation of the antibiotic therapy, vancomycin 500 mg every 6 h i.v. was added. The study population was at serious risk of a poor outcome, since 67% of the patients had leukemia or lymphoma, 19% of the febrile events occurred after autologous bone marrow or blood stem cell transplantation, the median total duration of neutropenia was 16 days, and the median neutrophil count at study inclusion was 0.09 × 109/l. The two patient groups were comparable in terms of risk factors. Bacteremia was found in 37%, other microscopically documented infections in 16%, and clinically documented infections in 26% of the febrile episodes. Most (96) febrile episodes were evaluable for response. No significant difference was found between piperacillin/ tazobactam and ceftazidime plus amikacin in terms of success rate (81% versus 83%), empirical addition of vancomycin (42% versus 38%), median time to fever defervescence (3.3 versus 2.9 days) or median duration of antibiotic therapy (7.2 versus 7.4 days). No patient died from the infection. Both antibiotic regimens were well tolerated, the study treatment being stopped only in 1 patient because of toxicity (cutaneous allergy to piperacillin/tazobactam). On the basis of the 107 febrile events encountered, we conclude that piperacillin/tazobactam is a safe and effective monotherapy. To define the definitive value of piperacillin/ tazobactam as a monotherapy for febrile neutropenic patients a large randomized trial is warranted.