In a previous phase I study, it was concluded that tolerable doses of busulfan (BU), cyclophosphamide (CY) and total body irradiation (TBI) were 8 mg/kg, 60 mg/kg and 12.0 Gy, respectively, for autologous marrow transplant recipients. In an attempt to decrease the variability of BU steady-state concentration (Css) following oral dosing, a BU dose escalation study based on targeted plasma levels was performed in patients receiving autologous transplants for AML or syngeneic transplants for CML. In this study, the BU dose was adjusted up or down based on observed plasma concentration. All patients received a fixed dose of CY 60 mg/kg and TBI of 12 Gy. The first dose level evaluated was 8.6 mg/kg with a target BU Css of 511 ng/ml. Eight patients were entered at this level and the median BU Css achieved was 441 (range 253-566). One of eight patients developed grade 3-4 regimen-related toxicities (RRT). The oral dose of BU for dose level II was 10.6 mg/kg with a target Css of 632 ng/ml. Six patients were entered at this level and median BU Css achieved was 642 (range 566-674). One of six patients developed grade 3-4 RRT. The oral dose for dose level III was 12.6 mg/kg with a target BU Css of 754 ng/ml. Five patients with AML were entered at this dose level and the median plasma BU Css was 733 ng/ml (682-900). Two of five (40%) patients at dose level III developed grade 3-4 RRT which was considered excessive making dose level II the MTD. This study showed that targeted BU Css can reliably be achieved with a bias of -5.23% and mean absolute error of 11.3%. Overall, targeting made a -32.5% to 158.3% change in plasma BU Css as compared to expected BU Css based on first dose pharmacokinetics if targeting were not performed in this study. Thus, targeting avoided much of the variability in BU Css seen in other studies and appears to have allowed for an increase in oral dosing from 8 mg/kg to 10.6 mg/kg. Despite achieving higher and more uniform BU Css, there was no apparent effect on relapse or survival, although the number of patients evaluated was small.