Résumé : Possible influences of donor (D)/recipient (R) HLA-DR antigen disparities on graft survival were studied in 310 renal cadaver transplantations performed at one center from January 1980 to July 1988. For each DR1 to DR7 specificity, four combinations were considered according to the presence (pos) or the absence (neg) of the antigen in the D and in the R. Logistic regression analysis was used to study graft survival during two postoperative periods: A (month 1 to 3) and B (month 4 to 24). The following combinations exerted a beneficial effect on survival: DR5 and DR7 Dpos/Rneg for period A, DR4 Dpos/Rneg and DR5 Dneg/Rpos for period B. In contrast, the following combinations appeared detrimental: DRw6 and DR7 Dneg/Rpos for period A, DR1 and DR2 Dpos/Rneg, DR2 and DRw6 Dneg/Rpos for period B. Graft survival at 2 years was similar (83%) in HLA-DR-identical pairs and in recipients with beneficial and no detrimental HLA-DR disparities, contrasting with poorer outcome in patients with either mixed beneficial and detrimental HLA-DR disparities (67%) or detrimental (and no beneficial) HLA-DR disparities (55%). The use of relevant D/R HLA-DR disparities for predicting graft outcome could yield results similar to those currently reported with the best HLA-DR-matched pairs, thus providing an alternative for recipient selection in cadaver kidney transplantation.