The impact of donor-recipient DPA1 and DPB1 matching was examined in 122 unrelated bone marrow transplant pairs. All pairs were serologically marched ar the rime of transplantation for HLA class I and II and a majority also DRB1 allele matched. Retrospective A, B, C, DRD1, DQA1, DQB1 in addition to DPA1 and DPB1 allele matching was performed by molecular techniques. The percentage of pairs that were allele matched was as follows; HLA-A = 91% (n = 80), HLA-B = 94%: (n = 80), HLA-C = 78% (n = 80), HLA-DRB1 = 96% (n = 122), HLA-DQA1 = 99% (n = 80), HLA-DQB1 = 92%, (n = 122). 92 recipient/donor pairs with informative clinical data were available for analysis. DPA1 identity (no incompatibility in either direction) was observed in 57% and DPA1 compatibility in 76% of pairs with no apparent beneficial effect of matching on patient survival or Graft Versus Host Disease (GVHD). DPB1 identity was observed in 11% and compatibility in 27% of pairs. A significant improvement in patient survival was observed in DPB1 marched compared ro one DPB1 mismatch (P < 0.01) and combined one and two DPB1 mismatched transplants (P = 0.03). This beneficial effect remained when allele mismatches at HLA-A, B, C, DRB1, DQA1, DQB1 were excluded (p = 0.05, p = 0.03, respectively). There was a significant association of increased frequency of severe GVHD (grades III-IV) compared to mild GVHD (grades I-II) with DPB1 mismatched transplants compared to DPB1 marched tr;transplants (p = 0.04). In DPB1 mismatched transplants an association between en patient survival and matching for individual DPB1 polymorphic regions was not observed; however in the HLA-A, B, DRB1, DQA1, DQB1 allele matched transplants a non significant increase in the frequency of Grade IV GVHD was observed in recipients who were negative compared ro those who were positive for DPB1 alleles coding for glutamic acid at position 69. (C) American Society for Histocompatibility and Immunogenetics, 1999. Published by Elsevier Science Inc.