Administration of combination chemotherapy to children with metastatic neuroblastoma induces profound myelosuppression resulting in chemotherapy treatment delays and febrile neutropenic episodes. The objective of this randomised multicentre study was to assess the incidence, duration and severity of neutropenia when filgrastim is added to induction chemotherapy administered to patients with metastatic neuroblastoma. In this study, 59 patients with metastatic neuroblastoma were randomised to receive chemotherapy (control group, n = 28) or chemotherapy plus filgrastim (filgrastim group, n = 31). Chemotherapy consisted of four cycles of cyclophosphamide, vincristine and doxorubicin (CADO) alternating at 21-day intervals with cisplatin and etoposide (CDDP-VP16). Filgrastim was administered subcutaneously at a dose of 5 micrograms/kg/day from day 7 for up to 14 days. The incidence of neutropenia (absolute neutrophil count [ANC] < 0.5 x 10(9)/l) in the filgrastim group was not reduced after the first CADO course. However, significant reductions were observed after courses 2, 3 and 4. The duration of neutropenia and of intravenous antibiotic use were significantly reduced in the filgrastim group over the whole study period (9 days versus 26 days, P < 0.001; 12 days versus 20 days, P = 0.04, respectively). However, the duration of hospitalisation and the incidence of febrile neutropenia were not significantly reduced. Compliance to treatment was good and the ability to administer chemotherapy without treatment delays was significantly better in the filgrastim group (P < 0.05). Event-free survival was greater in the filgrastim than in the control group (2.4 years versus 1.3 years; P = 0.072). Filgrastim is a beneficial adjunct to combination induction chemotherapy used in the treatment of metastatic neuroblastoma.