As the service life increases, the bridge performance deteriorates continuously, and the damage to structural components intensifies and spreads to adjacent components. The process of multiple independent updates of finite element models based on random search algorithms fails to reflect the dynamic variation of bridge service performance. The present paper puts forward a dynamic calibration method for bridge finite element models considering structural performance degradation based on the Cuckoo Search Algorithm (CS algorithm). Firstly, the set of parameters to be updated, which corresponds to the code of the CS algorithm, is divided into three groups: diseased components, adjacent components to diseased components, and other components. Subsequently, a range of strategies are employed to initiate each group. The upper limit of the parameters to be updated is set to the previous update result. In conclusion, the search range for each group of codes is limited by setting different control factors for step sizes. At the end of the paper, a single-span truss bridge was used to verify the effectiveness of the proposed method. The research results indicate that the dynamic calibration method for finite element models based on the CS algorithm can dynamically calibrate the finite element model of bridges. The dynamically calibrated finite element model can describe the time-varying behavior of the service performance of bridges as the damage deepens and spreads to adjacent components.