Abstract: The fine-scale investigation and evaluation of large-scale geological disasters are gradually being carried out. However, there are discrepancies between the existing susceptibility evaluation results and actual work. Exploring how to accurately obtain the susceptibility results of regional geological disasters is worth investigating. Taking Shifang City as an example, this study conducted field investigations based on slope units and continuously revised them. Based on the survey results, ten evaluation factors were selected through principal component analysis, correlation analysis and multicollinearity analysis. The information-logistic regression model is used to compare and analyze the susceptibility evaluation results of grid units and slope units. Finally, the susceptibility results of slope units were revised based on field survey data. The main conclusions are as follows: (1) The overall scale of geological disasters in Shifang City is relatively small, and susceptibility is mainly influenced by curvature, vegetation coverage, and roads. (2) The rationality and accuracy of grid units (AUC = 0.876 ) are higher than those of the slope unit. However, the results of grid units are fragmented and difficult to apply, while slope units have more high susceptibility areas and poor accuracy (AUC = 0.825). (3) Based on the field investigation, the susceptibility zoning of slope unit is revised. The proportions of high susceptibility, medium susceptibility, and low/non-susceptibility areas are 13.48%, 15.31%, and 71.21%, respectively, reducing the difficulty of control and matching the accuracy of field investigation results. The research results and evaluation process can guide local risk control work and provide references for similar studies.