Abstract: As a classical method of physical parameter identification, Extended Kalman filtering (EKF) algorithm is widely used in structural damage detection. However, due to the complexity of the state equations, most applications focus on the shear structure model with translational degrees of freedom considered only, and the comprehensive excitation and response signals are generally needed. Recognizing that the rotational degrees of freedom can be eliminated by the static condensation method, this paper proposes the dynamic condensation method considering Rayleigh damping, in order to establish the equivalent and simplified model for continuous girder bridges. According to the requirements of bridge inspection and health monitoring, the online and convenient damage detection method based on Extended Kalman filtering is presented, and the free vibration is generated by one hammer action, so that the deficiency of the traditional method needing complex excitation information is overcame. Taking a three-span continuous girder bridge as an example, the corresponding stiffness, the damage location and degree and the damping parameter were identified accurately by the proposed method, showing that the proposed method, with fast convergence speed, is feasible for the case of dynamic signal with high noise-signal ratio and practical engineering.