Seventy-six consecutive neutropenic patients with hematologic malignancies, admitted to the "Department of Hematology" of Rome between March and September 1986, were randomly assigned to receive either piperacillin (300 mg/kg in four divided doses) or ceftazidime (100 mg/kg in four divided doses) plus amikacin (15 mg/kg in two divided doses) whenever they developed a febrile episode (temperature greater than 38 degrees C thrice over 12 hours, not related to drugs or transfusions, or else temperature greater than 38.5 degrees C). After 72 hours of antibiotic therapy, in case of persistent fever, piperacillin or ceftazidime was added to the ceftazidime + amikacin or piperacillin + amikacin combination, respectively. The antibiotic treatment was, however, modified according to in vitro susceptibility if a positive culture was present. Success without regimen modification was observed in both antibiotic combinations in 52.6% of cases. Considering the empiric cross of antibiotics, the response rate reached 78%. Neither toxicity nor side effects were observed in the reported groups. Considering blood isolates, we observed a greater incidence of gram-positive organisms compared with gram-negatives (28 cases vs 5 cases, 84.7% vs 15.3% respectively). Fungal infections were documented in four cases, two in each group. Even though no statistical difference was found between the two groups as far as patients not responding to the first antibiotic combination are concerned, piperacillin seems to have had more efficacy (twelve patients responding to the addition of piperacillin vs seven patients responding to the addition of ceftazidime). Piperacillin + amikacin seems to be as effective as ceftazidime + amikacin in the empirical therapy of febrile episodes in neutropenic patients.