Abstract: Bayesian inversion approaches are usually used to quantify the significant uncertainty in the inversion results of the Multichannel Analysis of Surface Waves (MASW) dispersion curve. Due to the high-dimensional nonlinear geophysical model and extensive repeated forward calculations for the dispersion curve, Bayesian inference based on conventional numerical methods is highly time-consuming and less efficient. The paper proposes a probabilistic inversion approach for dispersion curves based on the Approximate Bayesian Computation (ABC) to achieve efficient identification of shear wave velocity (v_s) profiles. It can reduce the number of forward calculations for the dispersion curve, thereby greatly decreasing the computing time for Bayesian inference. Additionally, the subset simulation is used to improve the efficiency in solving high-dimensional Bayesian equations. Furthermore, the proposed approach can determine the most probable posterior value of v_s profiles, addressing a shortcoming of traditional ABC methods. The proposed approach is demonstrated through both virtual and real-life sites. Results show that the proposed approach is reasonable and highly efficient. This study provides valuable insights for conducting rapid probabilistic analysis of geophysical data.