Abstract Background In Russia, each transplant center has its own separated waiting list. This significantly reduces the ability to select the optimal recipient in terms of tissue compatibility. Aim to perform a multivariate analysis of the effect of the number of epitope mismatches on transplant survival. Method An observational retrospective cohort study was performed, which included 824 adult recipients. All patients underwent transplantation of a cadaveric kidney compatible by blood group. The endpoint was considered transplant loss. The death-censored graft survival was evaluated. The calculation of the number of epitope mismatches (EpMM) was carried out using information from publicly available resources about the population frequency of haplotypes and the repertoire of epitopes with confirmed immunogenicity. All possible combinations of the donor and recipient genotypes were compiled, and the probability of each combination was calculated. Then, for each combination with nonzero probability the number of donor epitopes absent in the recipient was determined. After that, the weighted average EpMM was calculated, where the weighted coefficient was the normalized probability of occurrence of each combination. Results All recipient donor pairs had HLA incompatibilities (HLA MM): 1.9% - 1 HLA MM, 6.7% - 2 HLA MM, 29.9% - 3 HLA MM, 38.5% - 4 HLA MM, 18, 1% - 5 HLA MM, 4.9% - 6 HLA MM. EpMM increased with increase of the HLA MM count, but not linearly: 6 [interquartile range – IQR 4; 7] (from 2 to 7), 12 [IQR 7.74; 17.25] (from 4 to 20), 18 [IQR 14; 22] (from 8 to 28), 24 [IQR 20; 30] (from 10 to 33), 30.5 [IQR 25; 37] (from 13 to 35) and 36 [IQR 26.5; 44.5] (from 15 to 40) for 1, 2, 3, 4, 5, and 6 HLA MM, respectively. Significant risk factors for graft loss in the adjusted multivariate model were: HLA MM (1 – HR 1; 2– HR 1.24 [95% confidence interval - 95% CI 0.7; 2.15], p = 0.344; 3– HR 1, 48 [95% CI 0.86; 2.33], p = 0.251; 4– HR 1.88 [95% CI 1.32; 2.52], p <0.001; 5– HR 2.41 [95 % CI 2; 2.93], p <0.001; 6– HR 2.98 [95% CI 2.59; 3.46], p <0.001) p <0.001; duration of conservation HR 1.08 per hour [95% CI 1.02; 1.16], p = 0.01; panel-reactive antibodies (PRA) HR 1.24 per every 10% [95% CI 1.06; 1.58], p = 0.01; but not the type of donor, age and gender of the recipient. When included EpMM in the model, significant risk factors were: EpMM (<10– HR 1; 10-19– HR 1.71 [95% CI 1.09; 2.49], p = 0.021; 20-29– HR 2,11 [95% CI 1.59; 2.68], p <0.001; 30-39– HR 2.4 [95% CI 1.96; 2.86], p <0.001; 40-49– HR 2, 59 [95% CI 2.17; 3.04], p <0.001; ≥50– HR 2.71 [95% CI 2.31; 3.15], p <0.001) p <0.001; PRA HR 1.18 for every 10% [95% CI 1.09; 1.5], p = 0.007; but not the duration of conservation, donor type, age and gender of the recipient. However, HLA MM was no longer a significant risk factor for HR graft loss 1.19 [95% CI 0.88; 1.55], p = 0.451. Increasing the EpMM from <10 to 10-19 significantly increases the risk of graft loss. This is an important aspect in the context of our study, since the number of EpMM 10-19 can correspond to 2-6 HLA MM. Validation of this method of assessing immunological load on a larger sample of patients is required. The list of immunogenic epitopes is constantly updated. The accumulation of this knowledge can improve results of transplantation. Conclusion The number of epitope mismatches remains an important factor in long-term transplant survival even when adjusted for other risk factors: the number of HLA mismatches, the donor type and the level of presensitization. This determines the possibility of using this method as an adjuvant in the donor-recipient pair selection. At the same time, there is a theoretical possibility of using this method as an alternative to the traditional assessment of tissue compatibility of a donor and a recipient.