A prospective randomized trial was performed to compare the efficacy of a regimen containing a glycopeptide versus one containing a beta-lactamase inhibitor in the treatment of febrile episodes in neutropenic patients. Fifty-eight patients received piperacillin/amikacin/teicoplanin (group 1) and 56 received piperacillin/amikacin/tazobactam (group 2). In the case of persistence of fever without microbiological documentation of the cause, teicoplanin was also given empirically in group 2 on day 4, and amphotericin B in both groups on day 6. In 114 evaluable febrile episodes, the rate of success without modification of therapy was 60% in patients on piperacillin/amikacin/teicoplanin and 41% in patients on piperacillin/amikacin/tazobactam (p < 0.03). Eleven of 34 patients in the latter group who failed to improve eventually responded upon addition of teicoplanin. Ten and nine patients in group 1 and group 2 respectively required the addition of amphotericin B for definite improvement. There were 14 episodes of gram-positive septicemia in each group: the response rate was 100% in group 1 and 43% in group 2. Three episodes of gram-negative breakthrough septicemia occurred in group 1 versus no cases in group 2 (p = 0.1). Three deaths occurred in each group. Piperacillin/amikacin/tazobactam may be as efficacious as piperacillin/amikacin/teicoplanin in the treatment of febrile neutropenic patients provided the regimen is modified (usually by addition of teicoplanin) in unresponsive cases.