Abstract: As the key indicator of the nuclear design, the computational accuracy of the core power distribution is very important for the evaluation of the economy and safety of nuclear power plants. As the first nuclear power software developed on self-reliance in China, the computational accuracy and applicability of NESTOR is the foundation for its application. Based on the random sampling statistical analysis (RSSA) method and deviation transmission idea, the uncertainty of core power distribution was obtained by combining two independent uncertainties resulting from the analysis of the uncertainty of physical model and the uncertainty of the change of parameters. The results indicate that the RSSA is feasible in the uncertainty analysis of nuclear design. In addition, in the analysis of the uncertainty of physical model, the core power distribution was decomposed into 2 parts, including the uncertainty of detailed power distribution in an assembly and the uncertainty of assembly power. As a result, the uncertainty of radial power distribution caused by physical model was ±3.653%, the uncertainty of radial power distribution caused by the change of parameters was ±0.964%, and the final uncertainty of radial power distribution was ±3.778% under the condition of 95% confidence coefficient and 95% probability that computed through the deviation transmission idea. The computational accuracy is as high as the engineering design software, and it lays foundation for the application and verification of NESTOR.