The efficacy and toxicity of teicoplanin and vancomycin in the initial empirical antibiotic regimen in febrile, neutropenic patients with hematologic malignancies were compared in a prospective, randomized, unblinded, multicenter trial in the setting of 29 hematologic units in tertiary-care or university hospitals. A total of 635 consecutive febrile patients with hematologic malignancies and chemotherapy-induced neutropenia were randomly assigned to receive intravenously amikacin plus ceftazidime plus either teicoplanin at 6 mg/kg of body weight once daily or vancomycin at 1 g twice daily. An efficacy analysis was done for 527 evaluable patients: 275 treated with teicoplanin and 252 treated with vancomycin. Overall, successful outcomes were recorded for 78% of patients who received teicoplanin and 75% of those who were randomized to vancomycin (difference, 3%; 95% confidence interval [CI], -10 to 4%; P = 0.33). A total of 102 patients presented with primary, single-agent, gram-positive bacteremia. Coagulase-negative staphylococci accounted for 42%, Staphylococcus aureus accounted for 27%, and streptococci accounted for 21% of all gram-positive blood isolates. The overall responses to therapy of gram-positive bacteremias were 92 and 87% for teicoplanin and vancomycin, respectively (difference, 5%; CI, -17 to 6%; P = 0.22). Side effects, mainly represented by skin rash, occurred in 3.2 and 8% of teicoplanin- and vancomycin-treated patients, respectively (difference, -4.8%; CI, 0.7 to 8%; P = 0.03); the rate of nephrotoxicity was 1.4 and 0.8% for the teicoplanin and vancomycin groups, respectively (difference, 0.6%; CI, -2 to 1%; P = 0.68). Further infections were caused by gram-positive organisms in two patients (0.7%) treated with teicoplanin and one patient (0.4%) who received vancomycin (difference, 0.3%; CI, -0.9 to 1.0%; P = 0.53). Overall mortalities were 8.5 and 11% for teicoplanin- and vancomycin-treated patients, respectively (difference, -2.5%; CI, - 2 to 7%; P = 0.43); death was caused by primary gram-positive infections in three patients (1%) in each treatment group. When used for initial empirical antibiotic therapy in febrile, neutropenic patients, teicoplanin was at least as efficacious as vancomycin, but it was associated with fewer side effects.