The low porosity and low permeability of tight sandstone pose considerable challenges on characterizing the reservoir fluid. Moreover, the geophysical characteristics of the tight sandstone are usually similar to the surrounding rocks, leading to small impedance contrast, compounding the difficulty of reservoir parameters inversion. Amplitude‐versus‐offset technology is an important tool to invert the reservoir parameters from seismic data. The existing amplitude‐versus‐offset inversion methods are challenged to predict the tight sandstone reservoir parameters for the reasons above. In this paper, we propose a new workflow based on amplitude‐versus‐offset attributes to characterize the tight sandstone gas reservoirs in the Xihu Depression, East China Sea. The proposed workflow utilizes the Backus average method and Gassmann's equation to construct the rock physics model. Based on the analysis of the anisotropic model, we introduce Rüger's anisotropic equation for amplitude‐versus‐offset forward modelling. We compare the amplitude‐versus‐offset gathers with the real data and adjust the reservoir parameters. Through the sensitivity analysis, we select the gas saturation and porosity as the inversion parameters. The relationship between the reservoir parameters and amplitude‐versus‐offset attributes is established by fitting the fifth‐order polynomials. We use Shuey's equation and the least squares inversion method to calculate the amplitude‐versus‐offset attributes (intercept and gradient). For real data, we introduce the Toeplitz‐sparse matrix factorization method to calculate the reflectivity of the amplitude‐versus‐offset gathers. Finally, we proposed a multi‐well joint inversion approach to invert the gas saturation and porosity. The application results demonstrate the effectiveness of the proposed workflow for the characterization of the gas reservoir in tight sandstone for the describedcase.