Cefmenoxime plus amikacin was compared in a prospective randomized trial with our standard regimen of piperacillin plus amikacin as an empiric therapy for fever in patients with granulocytopenia. Initial profound granulocytopenia (fewer than 100/mm3 mature granulocytes) was present in approximately 45% of the patients in trial of both treatment groups. Of 53 microbiologically and clinically documented infections treated with piperacillin plus amikacin, 36 (68%) showed improvement. Of 48 microbiologically and clinically documented infections treated with cefmenoxime plus amikacin, 23 (48%) showed improvement. The response rate for gram-negative infections treated with cefmenoxime plus amikacin was lower than that for infections treated with piperacillin plus amikacin. Toxicity was minimal, with an equivalent incidence of skin rash, diarrhea and hepatic dysfunction. Although clinical efficacy of the combination of piperacillin plus amikacin may be superior to cefmenoxime plus amikacin therapy, this study demonstrated no statistically significant differences.