The pharmacokinetics of an intravenous bolus dose of glycosylated recombinant human G-CSF (rhG-CSF) was examined in 15 patients with various hematological disorders and 3 normal volunteers. The elimination half-life of rhG-CSF varied with the disorder. The half-life of an initial dose of rhG-CSF (2 micrograms/weight kg) was significantly prolonged in patients with aplastic anemia (2.7 +/- 0.3 h, n = 3) and myelodysplastic syndrome-refractory anemia (2.0 +/- 0.3 h, n = 3) when compared with those in normal controls (0.9 +/- 0.5 h, n = 3). In contrast, in patients with acute myelogenous leukemia which was overt leukemia from myelodysplastic syndrome-refractory anemia with excess of blasts in transformation, the half-life was shortened after chemotherapy (0.2 +/- 0.1 h, n = 3). The half-life of rhG-CSF in 2 patients with acute lymphoblastic leukemia in complete remission was prolonged (2.0 and 2.7 h) at the time of marrow-suppression after chemotherapy and then shortened (0.5, 1.0 h, respectively) in the recovery phase. The half-life of rhG-CSF was very weakly, inversely correlated with absolute neutrophil count in blood (n = 24, r2 = 0.32, P < 0.01), and was inversely correlated with the absolute count of bone-marrow myeloid cells (nucleated cell count in bone-marrow aspirates x the percentage of myeloid cells/100) of patients with aplastic anemia and myelodysplastic syndrome-refractory anemia (n = 12, r2 = 0.63, P = 0.002). These results suggest that the half-life of intravenously administered rhG-CSF (2 micrograms/kg) reflects the size of the myeloid cell compartment in vivo, and support the hypothesis that receptor-mediated consumption mainly accounts for the clearance of exogenous G-CSF.