The reliability analysis approach of PMS (phased-mission system) based on Markov process encounters the state-explosion problem for large systems. To avoid the state-explosion problem, the PMS reliability model using hierarchical method was proposed. The hierarchical method uses a low-level continuous time Markov chain and an upper-level binary decision diagrams. The isomorphic nodes consolidation strategy and the redundant nodes removal strategy were given to construct reduced hierarchical model. The above nodes compressing strategies can improve the efficiency of model generation and storage. A recursive computing algorithm solving the hierarchical model was designed on the basis of the PMS components ordering, and its computational complexity was linear with the number of nodes in model. Finally, by a case study, the number of nodes in the original model was compared with the proposed model using nodes compressing strategy, also the mission reliability in the hierarchical model was compared with it in the Markov model. Results show the correctness and effectiveness of the reduced hierarchical model.