Abstract A method of analysis of HL-A compatibility in relation to the prognosis of kidney transplantation is described. It is based on the calculation of the probability of incompatibilities when the gene products of one or both of the defined HL-A loci are unknown in the donor, and on the assumption that cross-reacting antigens are not incompatible antigens. All of the 25 kidney grafts between HL-A genetically identical sibs have survived (P<0.001 when compared with HL-A genetically semi-identical grafts). Grafts between sibs sharing only one haplotype or between parents and children show a better prognosis when there is less than one calculated incompatibility. Among the 74 unrelated pairs, 70% of the matched pairs (less than two calculated incompatibilities) have functioned (P<0.01 when compared with mismatched pairs with 39% survival).