Forty-six patients receiving frozen bone allografts, preoperatively tissue typed for human leukocyte antigen and ABO antigens, were radiographically evaluated according to the Musculoskeletal Tumor Society scoring system at a mean followup of 55 months. Patients who matched for 1 or 2 Class I human leukocyte antigens with the donor scored higher than patients totally mismatched, but differences were not significant. Matching for Class II human leukocyte antigen and ABO antigens seemed not to influence radiographic outcome of allografts. In sixteen patients histologic specimens were obtained. Five of 16 patients who showed histologic parameters of an immune response scored significantly lower than those who did not. Processed frozen bone allografts, because of their lack of viable donor cells, most likely trigger an indirect pathway of alloantigen recognition in the recipient. This type of recognition may generate in the recipient either a chronic type of rejection or an immunologic state of tolerance to grafted antigens that cannot be measured with human leukocyte antigen blood tests. This may explain difficulties in correlating human leukocyte antigen mismatches between the donor and recipient with frozen bone allograft performances.