Abstract: Identifying modal parameters via processing vibration signals is one of the mainstream approaches for structural health monitoring and damage diagnosis. The processing approaches based on Fourier analysis are not able to process nonlinear and non-stationary signals. In addition, most of traditional identification methods suffer from low precision to identify damping. Therefore, a new approach is proposed for identifying modal parameters based on Hilbert-Huang transform (HHT) and natural excitation technique (NExT). First, the instantaneous characteristics of the original signal are extracted by means of empirical mode decomposition (EMD) and Hilbert transform (HT). Then, NExT and basic modal analysis theory are used to identify modal frequencies and modal damping ratios. Furthermore, the original acceleration record from the shaking table test of a 12-storey RC frame model is processed and modal parameters are identified by the proposed approach. And identification results are compared with the results from other identification algorithms and finite element analysis. Comparison indicates that the proposed approach is reliable to identify modal frequencies. Although identification of modal damping ratios gets improved by comparison with half-power bandwidth method, it is still difficult to confirm the precision of the results.