The prognostic significance of a positive B and/or T cell crossmatch test before transplantation was analysed in 174 cadaver kidney transplants. Thirty-one B cell-positive crossmatches were observed. In 15 of these cases, T cell crossmatches were also found to be positive retrospectively (long incubation assays). In 16 patients with a positive B cell crossmatch, an anti-HLA-DR antibody was present, as judged by platelet absorption studies, whereas in the other patients the B cell activity of the sera was attributable to weak anti-HLA-A,B,C antibodies or to cold autoantibodies. The success rate at 3 months of positive B cell crossmatch transplants because of anti-HLA-A,B,C or anti-HLA-DR antibodies was 84% whereas that of grafts with negative B and T cells crossmatches was 79%. Patients with B cell autoantibodies had excellent graft survival. Serial serum samples from patients receiving transplants were tested against specific donor B and T cells kept frozen. B cell antibodies appeared or persisted in 38 of 44 patients tested. Many patients also had donor T cell antibodies. Platelet absorption studies showed that about one-half of the B cell antibodies were presumably anti-HLA-DR and one-half anti-HLA-A,B,C. Donor-specific anti-B cell antibodies were detected in 86% of the patients at the time of graft rejection but they were also detected in approximately 80% of patients with excellent graft survival. In these cases they were usually also reactive with the patient's own B cells. These results indicate that at least two types of donor-specific B and/or T cell antibodies must be distinguished, those directed against HLA (A,B,C, or DR) antigens possibly deleterious, and those against autologous B lymphocytes, possibly enhancing.