To rapidly assess the reliability of crankshaft and consider the impact of engine speed on the fatigue life of crankshaft, a prediction method for crankshaft reliability was proposed based on dynamic Bayesian networks. The multi-body dynamics simulations of the crankshaft connecting rod mechanism were first conducted to obtain time-domain force data under various loads. The fatigue lives of crankshaft at different speeds were calculated by utilizing fatigue damage accumulation theory. The dynamic Bayesian network model was then employed to dynamically estimate the level of crankshaft reliability based on parameters such as the operating time and failure probability of crankshaft at different speeds, therefore establishing the changing laws of crankshaft reliability over time. The results showed that the crankshaft reliability was 61.78% by the 10th year.