We compared fractionated total body irradiation (12 Gy)/cyclophosphamide (120 mg/kg) with busulfan (16 mg/kg)/cyclophosphamide (120 mg/kg) as preparative therapy in unrelated donor stem cell transplantation of CML patients. Fifty patients with CML (1.CP = 46; aP = 4) and a median age of 36 years (range 16-52) were enrolled in this sequential trial between 1994 and 1999. In both groups patients were well balanced with respect to age, disease status, stem cell source and CMV status. All patients received standard doses of cyclosporin A, methotrexate and anti-thymocyte globulin (ATG) as GVHD prophylaxis. No graft failures occurred in either group. The median day of leukocyte engraftment was earlier in the Bu/Cy than in the TBI/Cy group (day 15 vs 17; P = 0.006). The incidence of grade II-IV GVHD was 40% in the TBI/Cy and 36% in the Bu/Cy group, whereas severe grade III/IV GVHD was only observed in 12% of patients in both groups. The incidence of chronic GVHD (limited and extensive) at 1 year was higher in the Bu/Cy arm (65% vs 30%; P = 0.02). More toxicity grade I/II of the liver (88% vs 44%; P = 0.002) and more hemorrhagic cystitis (32% vs 8%; P = 0.02) were observed in the Bu/Cy regimen. Seven relapses in the TBI and no relapse in the Bu/Cy group were observed after a median follow-up of 44 and 15 months, respectively. The estimated 3 year OS and DFS was 72% (95% CI: 55-98%) and 58% (95% CI: 39-77%) in the TBI and 70% (95% CI: 51-89%) for DFS and OS in the Bu/Cy group. We conclude that the anti-leukemic effect of the Bu/Cy regimen seems to be at least as effective as the TBI/Cy combination in unrelated stem cell transplantation of CML patients, with no graft failures, but that it correlates with a higher incidence of liver toxicity, hemorrhagic cystitis and chronic GVHD. Longer follow-up is necessary to determine the late relapse rate and late toxicity.