Abstract: Probabilistic safety assessment (PSA), as a component of the nuclear power plant (NPP) license document, plays an important role in the license application and safety operation of nuclear power plants. Human reliability analysis (HRA), as a key element of PSA in NPP, has an important impact on PSA analysis results and risk insights. There are many HRA methods, among which the standardized plant analysis risk HRA (SPAR-H) method has been widely used in China because of its ease to use. However, the step change of analysis result, the rationality of performance shaping factors (PSFs) considered, and the ambiguity of the selection criteria of different levels of a PSF are also received more and more concern. Based on engineering practice experience, theoretical research findings, and interviews with a group NPP operators, this paper attempted to optimize the SPAR-H method in terms of PSF categories, grading granularity and evaluation criteria of each PSF, and multiple factor of each PSF level. The PSFs in the optimized SPAR-H method included 9 PSFs: time pressure, stress, complexity, experience/training, procedures, human-machine interface, fatigue, working conditions and team cooperation. At the same time, the grades, definitions and evaluation criteria of each PSF were refined and clarified to improve the implementability. In addition, the multiple factors of each PSF level were further optimized to overcome the step change problem of the current SPAR-H method, and examine the validity of the improved method with data from an international HRA empirical study. The results show that the new SPAR-H method after PSF being improved has certain reliability and engineering feasibility in perform the quantitative analysis of human reliability. The optimized SPAR-H method presented in this paper can provide reference for the development of HRA method and the engineering application of the new method.