We conducted a randomized multicenter study to compare the efficacy and safety of two antibiotic regimens (cefepime [2 g b.i.d.] plus amikacin or ceftazidime [2 g t.i.d.] plus amikacin) as first-line therapy for fever in patients with hematologic malignancies and neutropenia. A total of 353 patients were randomized according to a 2:1 (cefepime:ceftazidime) ratio. Two hundred-twelve patients in the cefepime group and 107 in the ceftazidime group (90% of all patients) were evaluable for efficacy. The polymorphonuclear neutrophil count was < 100/mm3 on enrollment for 70% of the patients. The mean duration of neutropenia was 26 days. The efficacy in both study arms was comparable, although a trend in favor of cefepime was seen in terms of therapeutic success (response rate, 27% vs. 21% for the ceftazidime group). The overall response rate after glycopeptides were added to the regimens was 60% for the cefepime group and 51% for the ceftazidime group; the bacterial eradication rates were 81% vs. 76%, respectively, and the rates of new bacterial infections were 14% vs. 18%, respectively. We conclude that the combination cefepime/amikacin is at least as effective as the reference regimen of ceftazidime/amikacin in this setting.