Abstract: The sampling method developed based on evaluated nuclear data aims to more accurately quantify uncertainties and adjust prior nuclear data using measurement results and calculation uncertainties. This approach assumes that nuclear data distributions follow a normal distribution, as suggested by the maximum entropy estimation principle. The evaluated nuclear data sampling code, NECP-SOUL, was utilized to produce random nuclear data samples of ENDF-6 format, and then the nuclear data processing code, NECP-Atlas, was used to generate the application libraries based on each ENDF-6 sample. For transport calculations, the Monte Carlo code NECP-MCX and the deterministic code NECP-SARAX were employed to compute prior effective multiplication factors, control rod worth, and sodium void reactivity coefficients. To analyze correlations and identify reactors with high similarity, the S_SD similarity coefficient (similarity coefficient modified by standard deviation) was proposed to correct the conventional Pearson similarity coefficient. The Bayes-Monte Carlo approach was used to integrate measurement results and associated uncertainties to find optimal posterior nuclear data. The squared Mahalanobis distance was employed to quantify deviations between calculation and experimental values, and assign weight for each sample based on these deviations. The Bayes-Monte Carlo framework was used to the assimilation of nuclear data for two application aspects. Firstly, the China Experimental Fast Reactor (CEFR) and ZPPR-9 were set as target reactor, and several devices similar with these two reactors respectively were selected to assimilate evaluated nuclear data. The posterior nuclear data makes the deviation between the calculated and experimental k_eff of CEFR reduced from −588 pcm to 34 pcm, and ZPPR-9 from −700 pcm to 59 pcm. Secondly, the nuclear data were assimilated using the experiments performed on CEFR, and the posterior nuclear data were used to simulate CEFR itself. In this context, the maximum absolute deviation between calculated and experimental k_eff of CEFR was reduced from −1 016 pcm to −26 pcm, with a corresponding reduction in uncertainty from 1061 pcm to 79 pcm. The results show that the sampling-based method for uncertainty quantification and nuclear data assimilation develop in this work is effective for practical engineering applications. The developed method can be extended to shielding calculation, criticality safety analysis, and so on.