Platelet transfusion refractoriness results in adverse outcomes and increased healthcare costs. Managing refractoriness due to HLA alloimmunization necessitates the use of HLA antigen matched platelets, but requires a large platelet donor pool, and does not guarantee full matching. We report the first ever randomized, double-blind, non-inferiority, cross-over trial comparing HLA epitope-matched (HEM) platelets with HLA standard antigen-matched (HSM) platelet transfusions. Eligibility criteria were alloimmunized, platelet refractory, thrombocytopenic patients with aplastic anemia, myelodysplastic syndrome or acute myeloid leukemia. HEM platelets were selected using HLAMatchMaker epitope (specifically eplet) matching. Patients received up to eight prophylactic HEM and HSM transfusions provided in random order. Primary outcome was one-hour post transfusion platelet count increment (PCI). 49 patients were randomized at 14 UK hospitals. For intention-to-treat, number of evaluable transfusions was 107 and 112 for HEM and HSM, respectively. Unadjusted mean (SD) PCI for HEM and HSM was 23.9 (15) and 23.5 (14.1) respectively (adjusted mean difference -0.1, 95% CI -2.9, 2.8). As the lower limit of 95% CI was not greater than pre-defined non-inferiority limit, HEM was declared non-inferior to HSM. There were no differences in secondary outcomes of platelet counts, transfusion requirements and bleeding events. Adequate one-hour PCI was more frequently observed with a mean number of 3.2 of epitope mismatches compared to 5.5 epitope mismatches for inadequate one-hour increments. For every additional one epitope mismatch, the likelihood of an adequate PCI decreased by 15%. Epitope matched platelets should be considered to support HLA alloimmunized patients. Funded by NHS Blood and Transplant, ISRCTN23996532.