The success of bone marrow transplantation relies on class I and class II HLA identity between donor and recipient but until now the impact of DP mismatches on primary mixed lymphocyte reaction (MLR I) responses has not yet been clearly established. HLA-DP typing has been performed by a RFLP method on 10 patients and their 22 selected potential donors (HLA-A, B, DR, DQ, Dw identical according to serological, RFLP and oligotyping methods). HLA-DP-matching was then correlated with MLR I responses obtained (1) in 22 HLA-A, B, DR, DQ, Dw identical donor-recipient pairs (patient series) and (2) in 30 HLA-A, B, DR, DQ, Dw identical donors pairs used as the control series. We showed that MLR I responses involving patients cells were always lower than those involving control cells (p less than 0.02 in case of two DP-mismatches and p less than 0.001 in the case of one DP-mismatch). Moreover, in the control series two or one DP mismatches influenced MLR I responses in 91% and 77% of the cases respectively; however, there was a greater scatter of response values which could be explained by the degree of sequence homology between the DP mismatched alleles. In cases with no DP mismatch, no proliferative response was observed. Overall, these results suggest that the conventional technical conditions of MLR I do not allow detection of mismatches other than the well known HLA specificities.