Should HLA antigens that were mismatched in a renal transplant that failed be avoided in subsequent transplants? There were 890 retransplantations reported to the UCLA International Kidney Transplant Registry between 1985 and 1993 that had been performed in the face of a repeat HLA incompatibility. The 1- and 3-year regraft survival rates were 67% and 55%, respectively, for these retransplants, compared with 73% and 60% for 3220 regrafts with no HLA-A, -B, or -DR antigens mismatched twice (P = 0.030). When the repeat HLA-mismatched antigens were examined by locus, there was no difference in regraft survival comparing patients with no repeat HLA incompatibilities with those mismatched twice for HLA-A or -B antigens only, but there was a significant long-term decrease in survival of patients mismatched twice for HLA-DR antigens. The 377 patients mismatched twice only for HLA-A or -B antigens had 1- and 3-year regraft survival rates of 67% and 59%, respectively, compared with 65% (P = 0.289) and 50% (P = 0.025) for 281 patients with HLA-DR repeat mismatches only. Repeat mismatches for a combination of HLA-A or -B and -DR antigens resulted in 65% and 44% 1- and 3-year regraft survival in 129 patients. The half-lives for retransplants with repeat HLA class I, II, and I and II incompatibilities were 8, 6, and 4 years, respectively (P = 0.005). The data do not support preemptive avoidance of repeat HLA-A or -B incompatibilities. The crossmatch test excludes relevant mismatches. Repeated HLA-DR incompatibilities are not excluded by crossmatch tests and have a deleterious effect on long-term regraft survival. HLA-DR antigens mismatched in a previous failed transplant should be avoided.