Landslides pose a significant threat to communities in mountainous regions, particularly in the Himalayan areas. This study focuses on assessing the predictive accuracy of Geographic Information System (GIS)-based and machine learning models for landslide susceptibility mapping in the central Nepal Himalaya. We employed the bivariate Frequency Ratio (FR) method and the widely used ensemble machine learning technique, Random Forest algorithm (RF), to map landslide susceptibility within the Dharche Rural Municipality, which is near the epicenter of the 2015 Gorkha Earthquake. Our study involved preparing fifteen input layers for landslide conditioning factors, including slope, aspect, plan curvature, elevation, topographic wetness index, lithology, land use/land cover, distance from faults, rivers, rainfall, normalized difference vegetation index (NDVI), and proximity to roads. We identified a total of 571 landslide locations, with 70% randomly selected for training and the remaining 30% for validation. Our results indicate that the RF model outperforms the FR model in predicting landslide susceptibility. The findings of our research have significant implications for disaster management in the central Nepal Himalaya. They offer valuable insights for formulating disaster management strategies, aiding in the mitigation of landslide impacts on communities and infrastructure. By pinpointing high-risk zones, authorities can prioritize preventive measures and allocate resources more effectively. Overall, our results enhance disaster management efforts and inform development planning in this vulnerable region.