Unrelated donor (URD) hematopoietic cell transplantation (HCT) can eradicate chronic myelogenous leukemia (CML). It has been postulated that greater donor-recipient histoincompatibility can augment the graft-versus-leukemia (GVL) effect. We previously reported similar, but not equivalent, outcomes of URD versus sibling donor HCT for CML using an older, less precise classification of HLA matching. Here, we used our recently refined HLA-matching classification, which is suitable for interpretation when complete allele-level typing is unavailable, to reanalyze outcomes of previous HCT for CML. We found that using our new matching criteria identifies substantially more frequent mismatching than older, less precise "6 of 6 antigen-matched" URD-HCT. Under the new criteria, only 37% of those previously deemed "HLA- matched" were HLA well matched, and 44% were partially matched. Using our refined matching criteria confirms the greater risk of graft failure in partially matched or mismatched URD-recipient pairs compared with either sibling or well-matched URD-recipient pairs. Acute and chronic graft-versus-host disease (aGVHD, cGVHD) are significantly more frequent with all levels of recategorized URD HLA matching. Importantly, overall survival (OS) and leukemia-free survival (LFS) remain significantly worse after URD-HCT at any matching level. No augmented GVL effect accompanied URD HLA mismatch. Compared with sibling donor transplants, we observed only marginally increased (not statistically significant) risks of relapse in well-matched, partially matched, and mismatched URD-HCT. These data confirm the applicability of revised HLA-matching scheme in analyzing retrospective data sets when fully informative, allele-level typing is unavailable. In this analysis, greater histoincompatibility can augment GVHD, but does not improve protection against relapse; thus the best donor remains the most closely matched donor.