Abstract: In landslide susceptibility evaluation, there is still no consensus on the issue of index weighting. How to effectively balance subjective and objective weights while reducing disputes over weighting and improving the scientificity and accuracy of model evaluation remain one of the key challenges. This study selected ten influencing factors, including rainfall, elevation, vegetation normalization index, and slope, and proposed an evaluation method for landslide susceptibility based on the TOPSIS-partial order set model to address the dependence on weight values in traditional evaluation approaches. In this method, only the relative order of index weights is required. By deriving the positive and negative ideal points and the cumulative transformation matrix, the calculated height value of each sample is taken as the susceptibility probability. This approach simplifies the computational process and enhances model stability and applicability. Using the landslide inventory data of Hanyuan County, Ya 'an City as a case study, the evaluation results of this model were compared with those of the SVM model. The results show that the TOPSIS–Poset model achieved a susceptibility evaluation accuracy (AUC) of 0.957, significantly higher than that of the SVM model (0.912). The susceptibility zoning results indicate that high- and very-high-susceptibility zones account for 21.67% and 18.28% of the total area, respectively, and that 76.37% of historical landslides are located within these two zones. These findings demonstrate that the susceptibility zoning map generated by the TOPSIS-partial model is consistent with the spatial distribution and development patterns of landslides, providing valuable reference for regional landslide susceptibility assessment.