Abstract: When the traditional reliability analysis method is used to solve complex multi-state system problems, the system cost is often increased, which causes high computational cost and low precision. Aiming at this situation, proposes a manufacturing system reliability analysis method based on Markov stochastic process method and universal generating function (UGF). Firstly, the Markov model of each unit of the manufacturing system is established by using the stochastic process method to solve the unit state probability. The UGF model of each unit is determined by Lz transformation, and then the UGF model of the whole system is obtained through compound operation, at the same time, the transfer density matrix of the system is optimized and solved by genetic algorithm, and the markov reliability model of the system is established. Combined with numerical examples, the feasibility and effectiveness of the proposed method are verified. The new method has reference significance for engineering applications.