Abstract: To improve the continuity of parallel production process, a parallel system preventive maintenance model (3M1B) considering buffer stocks allocation is proposed. The possibility of imperfect production of the equipment is considered, that is, when the equipment becomes incomplete, there is a certain probability to produce defective products. The strategy of optimal preventive maintenance and the buffer stocks allocation are devised by establishing the preventive maintenance model with the minimum production cost rate as the objective function and taking the buffer stocks and equipment operation cycle as the decision variables. For this maintenance model, the discrete iteration algorithm and genetic algorithm are used to solve the problem, and the advantages and disadvantages of the two algorithms are compared. The results show that the discrete iteration algorithm is better than the genetic algorithm for this model. At the end, the feasibility and effectiveness of the model are verified by a numerical example.