HLA incompatibility between bone marrow recipient and unrelated donor pairs is often associated with severe acute graft-versus-host disease following bone marrow transplantation. Due to the extensive polymorphism of HLA genes, finding genotypically identical pairs is a difficult challenge. Therefore, it is crucial to single out the relevance of each HLA gene and, within each sequence, the polymorphic positions that induce a T-cell response. Among HLA class II genes, the relevance of HLA-DPB1 in inducing graft-versus-host disease is still controversial. In this study, we selected 37 bone marrow transplant pairs on the basis of HLA class I A and B identity as determined by isoelectric focusing and of class II identity as determined by serology and by low-resolution genomic typing. We analyzed them for the possible relationship between frequency of cytotoxic T lymphocyte and helper T lymphocyte precursors (CTLp and HTLp, respectively) and genomically determined class II mismatches. Seventeen pairs had high numbers of both CTLp and HTLp. They were not further considered because of the difficulty in determining whether the T-cell response was induced by class I or class II mismatches. Of the remaining pairs with low CTLp and high HTLp, six had disparities at HLA-DRB1 and HLA-DPB1 genes, and 14 differed only at the HLA-DPB1 locus. Among the latter pairs, we found a correlation between HLA-DPB1 mismatches and HTLp frequency, thus suggesting that disparity at this locus influences the alloreactive T-cell response. When the HTLp frequency was correlated with each single mismatch found in the 14 pairs, it appeared that the nature of the amino acid at position beta69 played a relevant role in inducing alloreactivity.