Aiming at the problems of multi-dimensional small failure probability and implicit function in the reliability analysis for bridge crane structure, that is difficult to solve using traditional reliability analysis methods, the one based on dynamic Kriging surrogate model combined with important sampling method is proposed. Firstly, the dynamic Kriging surrogate model of the implicit function is constructed utilizing two learning functions, and then the most probable failure point (MPP) is obtained by using the improved first order second moment method combined with the Kriging surrogate model. Secondly, the importance sampling density function is constructed with the MPP as the central point. Finally, the reliability degree is calculated by the importance sampling method based on the established surrogate model. By introducing a new stop criterion combined with learning function, the number of finite element calls is reduced. Verified by an engineering case, the proposed method can well balance the model accuracy, result error and calculation cost.