Fifteen children (age 1.2 to 9.4 years) with advanced neuroblastoma were treated with myelosuppressive chemotherapy (cyclophosphamide, cisplatin, doxorubicin) followed by 5 (n = 5), 10 (n = 5), or 15 (n = 5) micrograms/kg recombinant granulocyte colony-stimulating factor (rG- CSF) subcutaneously (SC) once daily for 10 days, starting the day after chemotherapy. Serial serum samples obtained on days 1 and 10 were analyzed for G-CSF activity by a specific proliferation assay using NFS- 60 cells. G-CSF serum concentration-time data were best described by a one-compartment model, with zero-order absorption and first-order elimination. After SC injection, absorption was prolonged, with peak concentrations of G-CSF (3 to 117 ng/mL) being reached after 4 to 12 hours. The relatively slow absorption, with a mean elimination half- life of 5.8 hours on day 1 and 4.5 hours on day 10, provided measurable G-CSF concentrations for the entire 24-hour dosing interval in all patients at each dosage level. The median apparent clearance of G-CSF on day 10 was significantly higher than on day 1 (0.57 v 0.31 mL/min/kg, P = .02), and was positively correlated with the absolute neutrophil count (ANC) (r2 = .33, P = .003). Systemic exposure to G-CSF was dose-related, but interpatient pharmacokinetic variability yielded overlap in area under the concentration-time curve (AUC) at all three dosage levels. Stepwise regression analysis showed that G-CSF AUC could be predicted by a model that includes rG-CSF dosage and ANC on the day of administration (r2 = .82, P = .0001).