The double beta-lactam combination of moxalactam plus piperacillin was compared with the aminoglycoside-containing regimen of moxalactam plus amikacin in a prospective, randomized trial of empiric therapy for 302 febrile episodes in granulocytopenic cancer patients. The moxalactam/piperacillin regimen was found to be as effective as the moxalactam/amikacin regimen (70 percent overall responses); responses with moxalactam/piperacillin and moxalactam/amikacin were similar for microbiologically documented infections (24 of 37, 65 percent, versus 20 of 35, 57 percent), for the subgroup with bacteremias (19 of 32 versus 14 of 28), and for clinically documented infections (41 of 58, 71 percent, versus 40 of 48, 83 percent). Responses were similar also for bacteremia in patients with persistent, profound (less than 100/microliter) granulocytopenia. Among profoundly (less than 100/microliter) granulocytopenic patients with gram-negative bacteremia, an increase in the granulocyte count to more than 100/microliter during therapy and a peak bactericidal activity of 1:16 or more (the latter noted in seven of nine moxalactam/piperacillin trials and six of nine moxalactam/amikacin trials) correlated with a favorable clinical response in 85 percent (p less than or equal to 0.00003) and 92 percent (p less than or equal to 0.044), respectively. Although serious side effects were minimal with either regimen, the double beta-lactam combination was associated with significantly less frequent nephrotoxicity (two of 145 versus 12 of 130; p less than or equal to 0.003) and ototoxicity (none of 34 versus seven of 34; p less than or equal to 0.006). The double beta-lactam combination of moxalactam plus piperacillin was found to be as effective as moxalactam plus amikacin but to have significantly less nephro- and ototoxicity.