Three protocols of blood transfusion were evaluated in a canine model for (1) the strength and breadth of leukocytotoxin induction, (2) the induction of cell-mediated immunity against the blood donors, (3) haemagglutinin production, and (4) any effect on kidney graft survival. At the end of the transfusion schedule, each dog received a kidney graft and was given azathioprine and prednisolone postoperatively. All dogs were unrelated and blood donors were not used as kidney donors. All three transfusion protocols, comprising i.v. injections of blood twice weekly or every 2 weeks from one or three donors, induced unacceptably strong and broad leukocytotoxins. All transplants performed across a positive crossmatch failed to function. However, where a negative crossmatch was available, the trend of results was that the transfused dogs had better graft survival than nontransfused animals similarly treated with azathioprine and prednisolone. Only one dog produced haemagglutinins. Several animals had positive cell-mediated immunity against the blood donors, but the response was not strong and was frequently not sustained. The problems posed by blood transfusions in clinical renal transplantation are complex, and have been a subject of debate from the time of the earliest transplants. A number of early studies comparing graft survival in patients, who had received few transfusions with those that had received many before transplantation, showed that graft outcome either was not different in the two groups (19) or that it might have been better in the patients given the most transfusions (7). That blood transfusions before transplantation could induce a state of specific unresponsiveness to a subsequent organ allograft was clearly demonstrated in the rat by a number of groups (10, 17, 18). The more exacting clinical comparison of a group of nontransfused patients with a transfused group was not made until 1974 when Opelz and Terasaki (21) published data showing that graft survival was substantially worse in nontransfused patients. Their work has been supported by further clinical reports (14, 15, 22, 27) and by experimental work in the rhesus monkey (26) and the dog (1). Although the clinical results from our unit show that nontransfused and transfused patients have essentially similar and excellent graft survival (P. J. Morris et al., in preparation), the retrospective studies from most, but not all, units suggest that blood transfusions before transplantation do lead to better graft survival in those patients that receive a transplant.The problem that arises in patients given transfusions before transplantation is that any beneficial effect in terms of graft protection for those that come to transplantation might be offset by the relegation to life-long dialysis of those that become highly sensitized (3, 24). The clinical problem, then, is the choice of a protocol for transfusion that is most likely to result in graft protection and least likely to result in the production of lymphocytotoxins to HLA-A, B, and C antigens. The balance between an overall harmful or beneficial effect of transfusions almost certainly depends on a number of variables, such as the number of different blood donors, the frequency of transfusion, the volume of blood transfused, the use of frozen blood, the age of the blood, the interval between the last transfusion and transplantation, the degree of histocompatibility between blood donor and recipient, the genotype of the recipient, and the degree of nonspecific immunosuppression induced by chronic renal failure. The experiments reported in this paper attempt to evaluate some of these factors in the dog.