Background: Hematopoietic stem cell transplantation (HSCT) cures leukemias, but comes at a cost of higher toxicity and non-relapse mortality (NRM). We previously showed that fractionating the busulfan (Bu) dose and starting it on day -13 instead of day -6, leads to lower toxicity and permits delivery of a myeloablative regimen, even to older patients. We designed a single institution randomized trial to validate the concept of fractionation and to test the hypothesis that further fractionation by starting chemotherapy on day -20 further reduces toxicity and lowers NRM without increasing relapse risk. Methods: Patients with AML or MDS, 18 to 70 years old, with a matched related or unrelated donor were eligible if they had adequate organ function. Patients were randomized to one of two arms: The shorter arm consisted of outpatient Bu 80mg/m² on days -13 and -12, while the longer arm consisted of outpatient Bu 80mg/m² on days -20 and -13. The rest of the regimen and GVHD prophylaxis were similar in both arms: On days -6 to -3, fludarabine 10mg/m², cladribine 10mg/m² and Bu dosed to achieve total Bu exposure of 20,000 ± 12% µMol-min for the whole course, including outpatient doses. Additionally, the last 33 patients received venetoclax from days -22 to -3 in both arms. The GVHD prophylaxis was tacrolimus and mini-methotrexate in the first 29 patients and tacrolimus and cyclophosphamide in the remainder. The only difference between the two arms was if busulfan was started on day -20 or not. Results: A total of 116 patients were randomized, 59 to the shorter day -13 arm and 57 to the longer day -20 arm. Median age was 59 in the day -13 arm and 57 in the -20 arm. 70.7% had AML. 41.4% were transplanted in first complete remission. The characteristics of the two groups were well-balanced. The median follow-up in the 67 surviving patients was 48.9 months. More patients in the day -13 group experienced grade ≥3 toxicities than in the day -20 group (93.2% vs. 73.7%, p = 0.005). The difference in toxicity led to a reduction of the 3-year NRM (19% vs. 7%, p = 0.067). There was no difference in relapse rate (p = 0.99) (Table). Conclusions: Further fractionating busulfan and moving just one dose of busulfan 1 week earlier reduces toxicities and may lower NRM, without increasing relapse rate. Clinical trial information: NCT02250937.