Human reliability analysis (HRA) is an expansion of man-machine engineering. It is also a new multidisciplinary based on behavioral science, cognitive science, information processing, system analysis and probability statistics in order to analyze, predict, reduce and prevent human errors. Firstly, the quantitative analysis model of HRA is proposed based on Markov process theory by using human error probability (HEP) and error correction cycle (ECC) as parameters. And human reliability evaluation criterion is built. Then, the HRA process considering error correction is proposed based on cognitive reliability and error analysis method (CREAM). Finally, according to the characteristics of armored vehicle system, common performance condition (CPC) in CREAM is improved. A reliability impact index is characterized by the overall contexts of tasks. Human reliability evaluation criterion of armored vehicle system is formulated. And the result of HRA is obtained based on the method presented in this paper. In addition, the relative weights are estimated by combining scale of 10/10—18/2 and analytical hierarchy process (AHP), and the triangular fuzzy number considering confidence factor and optimism index is adopted in order to reduce the subjectivity. The analysis results show that the method presented in this paper is reasonable and feasible. Meantime, the method can provide guidance for human reliability analysis of other weapon systems. © 2016, Shanghai Jiaotong University and Springer-Verlag Berlin Heidelberg.