A prospective randomized trial was performed to compare teicoplanin to vancomycin as part of the empirical antibiotic therapy of febrile neutropenic cancer patients. Fifty-three patients were randomized to receive ceftazidime (100 mg/kg daily every 8 h), amikacin (15 mg/kg daily every 8 h) and teicoplanin (6 mg/kg once a day) and 53 other patients received ceftazidime, amikacin (same dosages) and vancomycin (30 mg/kg/day every 6 h). In 99 evaluable episodes, the success rates were 54% for patients receiving teicoplanin and 52% for patients receiving vancomycin (p=0.76, 95% CI-18-23). The response rates were similar for patients with unexplained fever and for patients with documented infections. There were no differences in renal toxicity or cutaneous side effects between the two groups. The overall death rate was 18.9%, with 10 deaths in each group. The most important factor associated with death was the diagnosis of a fungal infection (p=0.001). Teicoplanin seems to be well tolerated and as effective as vancomycin in the empirical antibiotic therapy of fever in neutropenic cancer patients.