The optimal switching and condition-based maintenance decision of the cold/warm mixed standby system with periodic switching are studied. The system structure and switching operation and maintenance characteristics are analysed on the basis of system degradation modelling, and the off-line condition-based maintenance strategy with periodic switching and detection is formulated. Firstly, by analyzing the state transition characteristics of the system in the process of alternate use and maintenance of active and standby equipment, the model of probability distribution of the system state and the probability of each maintenance activity are derived. Then, an analytical optimization model is established to minimize the average cost rate in the finite time horizon, and the optimal switching period and maintenance threshold are determined. The model is solved using the genetic algorithm. Finally, the correctness and effectiveness of the strategy and model are verified by taking the stator cooling water pump system of a turbogenerator as an example, and the sensitivity analysis of the parameters is carried out. The results show that the proposed off-line condition-based maintenance strategy can effectively reduce the maintenance cost of the system.