PURPOSE: This study was designed to evaluate the safety and efficacy of early cessation of antibiotic treatment regardless of absolute neutrophil count in children with febrile neutropenia and no identifiable infectious source. METHODS: A prospective randomized clinical trial was performed in 93 episodes of fever in 37 neutropenic children with cancer, who were admitted to the Department of Pediatrics, Pusan National University Hospital from January 1996 to June 1997. RESULTS: Thirty-seven patients represented 93 febrile neutropenic episodes after chemotheraphy. Among them, 63 febrile episodes had no focus of fever, and there were 10 episodes of bacteremia. Of 63 febrile episodes who weren't identified focus of fever, only 47 febrile episodes had early cessation of antibiotics. During 47 febrile episodes, only 4 episodes (8.5%) had recurrent fever after early cessation of empirical antibiotic therapy. Between patients with recurrent fever and without recurrent fever, there were no significant difference in absolute neutrophil count (ANC) and CRP at admission, high body temperature before admission and duration of antibiotics. Recurrent rate of fever was high in patients who declined in absolute neutrophil count (ANC) and less than 25% increase in absolute phagocyte count (APC) at the stop of antibiotics. CONCLUSION: We suggest that early cessation of antibiotics regardless of absolute neutrophil count (ANC) is safe in the case of febrile neutropenia without an identifiable source of infection.