Abstract: A new parameter identification algorithm based on the state-space and wavelet transform is presented in this paper, which uses the system excitation and the response data. For an arbitrarily linear time-varying structure, the second-order vibration differential equations can be rewritten as first-order vibration differential equations using the state-space method. Both excitation and response signals are projected by the Daubechies wavelet scaling functions, and then the state-space equations of the time-varying dynamic system are transformed into simple linear equations using the orthogonality of the scaling functions. The time-varying equivalent state-space system matrices of the structures at each moment are then identified directly by solving the linear equations. The modal parameters are extracted via eigenvalue decomposition of the state-space system matrices and the time-varying stiffness and damping matrices can be determined by comparing the identified equivalent system matrices with the physical system matrices. A 2 degrees-of-freedom spring-mass-damping model with three kinds of time-varying cases (abruptly, smoothly and periodically) is investigated. Numerical results show that the proposed method is accurate and effective to identify the time-varying parameters.