In the design phase of a mine hoist’s main bearing， the reliability analysis of its random vibration cannot obtain complete probability information of the vibration acceleration response due to insufficient experimental samples. This paper proposes a new technical route for the reliability analysis of the main bearing of a mine hoist under incomplete probability information. This route in? cludes the dynamic model of a multi-scale coupling system for rolling element bearings， the probability density evolution of vibra? tion acceleration， stochastic process modeling of the probability density evolution path， the probability distribution of vibration pow? er spectral density， and the calculation of design reliability based on conditional probability. By using the collected condition data to drive the established dynamic model of multi-scale coupling system for rolling element bearings， the probability density evolution of vibration acceleration is carried out. Based on the probability density evolution and Karhunen-Loève expansion， a modeling ap? proach for the non-stationary random process of vibration acceleration of rolling element bearings is proposed. This approach ob? tains the twin data of the random sequence of vibration acceleration for rolling element bearings. The probability distribution of the vibration power spectral density for the main bearing of a mine hoist is studied， and the reliability index of the main bearing of a mine hoist within the service time is calculated.