To assess the efficacy of recombinant human stem cell factor (rHuSCF), 48 patients who had failed to mobilize >2.0 x 10(6) CD34+ cells/kg with granulocyte colony-stimulating factor (G-CSF) (10 microg/kg twice daily) with, or without, concomitant chemotherapy (G-CSF-based regimen), were remobilized with the addition of rHuSCF (20 microg/kg/day). In all, 18/48 (38%) achieved a total of >2.0 x 10(6) CD34+ cells/kg with the second rHuSCF-based mobilisation alone and 29/48 (60%) achieved a cumulative total of >2.0 x 10(6) CD34+ cells/kg following remobilization. Inclusion of chemotherapy in the mobilization regimen resulted in a higher yield of CD34+ cells/kg for both the initial G-CSF-based and subsequent rHuSCF-based regimens (0.90 vs 0.54, P < 0.01 and 2.36 vs 1.34, P < 0.01, respectively). The total peripheral blood stem cells PBSC collected from the G-CSF-based regimen, performance status, baseline platelet count and albumin were significantly associated with successful remobilization. Patients with multiple myeloma were also more likely to successfully remobilize. There was no threshold of total collected from the failed G-CSF-based regimen below which successful remobilization with the rHuSCF-based regimen was not possible. We therefore propose a predictive model [PBSC expected = 0.6+(G-CSF-based total collection)+2 (rHuSCF-based day 1 collection)] to calculate the cumulative total of PBSC expected following a maximum of five leukaphereses. This algorithm may permit the early identification of patients who are unlikely to achieve sufficient PBSC for transplantation and allow physicians to direct the resources involved in PBSC collection in a more appropriate and economical manner.