Abstract: The location of the critical slip surface is a very important issue in slope stability analysis and reinforcement design. In this study, an improved particle swarm optimization (VSPSO) algorithm is proposed to search for the critical slip surface based on particle variation (PV) and sequential quadratic programming (SQP). PV enhances the ability of PSO in jumping out of the local optimum, and SQP accelerates local search. The combination of PV and SQP greatly promotes the capacity of the original PSO in looking for the global optimum. An example with analytical solution is analyzed by the VSPSO, and the results demonstrate the accuracy and efficiency of the proposed model. Three typical examples from ACADS are then given, which are respectively homogeneous slope, multilayer soil slope and slope with weak layer. It is shown that results from the VSPSO are more approximate to the recommended values than those from the PSO. Furthermore, the VSPSO has a quite well robustness for the slopes with very complicated geometries and material properties.