Twenty-three children with congenital or acquired hematological disorders and 8 children with solid tumors received filgrastim at a dose of 5 micrograms/Kg by a daily 2-hour infusion following allogeneic (18 cases) or autologous (13 cases) bone marrow transplantation (group I). The results were compared with those of a disease, age and type of transplant matched cohort of 31 children treated in the same institution who did not receive the growth factor (group II). Filgrastim treatment was started within 24 hours of completion of the marrow infusion and lasted for 21 consecutive days or until the absolute neutrophil count reached 10 x 10(9)/l for 3 consecutive days. Twelve variables were evaluated prospectively in Group I and retrospectively in Group II. Myeloid reconstitution with peripheral granulocyte counts > 0.5 x 10(9)/L was achieved at a median time of 13 days in group I and of 14 days in group II (p = ns). Platelet recovery to > 50 x 10(9)/L was slower in group I (43 vs 30 days: p < .05). Median time to last platelet and red blood cell infusion was higher in group I (33 vs 18 days for platelets, p < .05; 45 vs 25 for red blood cells, p < .005). Filgrastim-treated children undergoing autologous BMT had fewer days of fever (6 vs 10 days, p < .05). There was no significant toxicity ascribable to filgrastim. Clinically and microbiologically documented infections, days of antibiotic therapy, duration of total parenteral nutrition and median time in hospital were similar in both groups. We conclude that in children undergoing autologous BMT for malignancies, filgrastim significantly reduced the number of febrile days. Similar benefits were not observed in children undergoing allogeneic BMT. Children receiving filgrastim experienced a delay in erythrocyte and platelet recovery. A prospective randomized study is required to better define the cost-benefit of filgrastim in children undergoing autologous or allogeneic BMT.