Abstract: As one of the key parts of aero-engine, the reliability of cylindrical roller bearings affects the stability of aero-engine to some extent. Therefore, it is very important to study the reliability of cylindrical roller bearings of aero-engine. However, the complex limit state functions often appear in the reliability analysis of cylindrical roller bearings, such as highly nonlinear functions, and even some limit state functions are implicitly given. At this time, it is an effective method to use surrogate model to approximate the limit state function of cylindrical roller bearings. In this paper, by combining with Kriging surrogate model and the active learning (active learning Kriging-Monte Carlo simulation) ALK-MCS algorithm, the reliability of an aero-engine’s cylindrical roller bearings is studied. Firstly, the three-dimensional model of a cylindrical roller bearing is established, then the finite element simulation is performed, and finally the ALK-MCS-based reliability analysis of the cylindrical roller bearing is carried out. The results show that as an active learning reliability method with high efficiency, ALK-MCS algorithm can significantly reduce the iteration. Simultaneously, the reliability result of the cylindrical roller bearing through ALK-MCS algorithm provides some guidance for the decision-maker in actual engineering.