<p>Slope units provide comprehensive geomorphological information and have been widely utilized in landslide susceptibility studies. However, studies on the impact of the quantity of slope unit divisions on landslide susceptibility assessment are relatively scarce. This study, taking Songpan County in China as a case, obtained five different quantities of slope units. We refined the initial fifteen factors through geographical detector and multicollinearity analysis methods and trained the data using three machine learning models. The best overall performance, was chosen for landslide susceptibility mapping. It analyzed the proportions of different landslide susceptibility categories under varying quantities of slope units and grid units, in conjunction with the disaster activity intensity index (R). The configuration with 20,470 slope units yielded the best results, having the smallest proportion of high and very-high susceptibility areas, at 15.5% and achieving the highest R value of 5.26. The results indicate that the quantity of slope units affects the contribution of landslide influencing factors. Compared to grid units, the q values of factors in different quantities of slope units changed, indirectly affecting the factor selection results. When using slope units for landslide susceptibility mapping, it is crucial to consider their quantity, as the quality of landslide susceptibility mapping does not improve with finer division of slope units, and evaluating landslide susceptibility with slope units does not necessarily yield better results than with grid units. Selecting an appropriate number of slope units can make the landslide susceptibility results more scientific and accurate.</p>