Abstract: A stochastic subspace system identification method is used to extract modal parameter of HUMEN suspension bridge in order to get accurate structural dynamic characteristics from ambient experiment. The results are essential for structural anti-wind, aseismatic and real time monitoring. With the response data of ambient vibration, this time domain identification method can extract structural modal parameters based on the state space model, avoiding conventional manual identification and iterative process. But the stabilization diagram must be used to determine the modal sequences. Comparing to FEM numerical analysis, the method proposed in this paper can identify ten modes below 0.5 Hz and reliable damping ratio. Therefore, it is concluded that the stochastic subspace system identification method is a powerful tool to acquire the characteristic parameters of long-span bridges.