Abstract: Sand slide slope are sudden, unpredictable and has great harm.Through field investigations around Lhasa, 12 representative data sets of sand slide slope disaster points were collected. By combining Bayesian network and Particle Swarm Optimization algorithm, and using algorithm update formula to make up for the short comings of the single algorithm, information entropy was introduced to analyze the weight of rainfall, slope, slope height and vegetation coverage in the algorithm.Then, the influence of various factors on the stability of sand-pass slope is analyzed and the stability of sand-pass slope is graded. Experiments proved that the method is effective and has certain reference value for the stability evaluation of sand slide slope.