Abstract: Big data transmission and storage cost caused by massive data collected and poor sparsity of actual vibration signals polluted by external environmental noise are critical problems in structural health monitoring. Thus, vibration responses reconstruction based on Bayesian compressed sensing combined with threshold method is proposed in this paper. The Bayesian compressed sensing theory is introduced in this paper to reduce the sampling, and the threshold method is used to enhance the sparsity of the collected signals to improve the accuracy of responses reconstruction. Then the fast hyper-parameter estimation method is utilized to estimate the sparse vectors to reconstruct the original signals with high accuracy and probability. The effectiveness and feasibility of the Bayesian compressed sensing vibration signal reconstruction algorithm combined with the threshold method were verified using ambient vibration responses from the Ji’an Bridge of Jiangxi Province. The reconstructed results indicate that the accuracy of vibration signal reconstruction using the proposed method is superior to that of Bayesian compressed sensing reconstruction without using threshold method.