The Shangri-La region in northwest Yunnan preserves extensive glacial moraines, which provide abundant loose material for landslides and debris flows. Mapping these moraines is therefore critical for susceptibility assessment. In this study, we used remote sensing interpretation to extract the spatial distribution of moraines. We then applied the Analytic Hierarchy Process (AHP) to eight factors—elevation, slope, curvature, vegetation cover, lithology, moraine distribution, rainfall, and human activities—and built a susceptibility model within a GIS framework. A Monte Carlo simulation was also conducted to test the sensitivity of the factor weights to potential uncertainties in expert judgments. The results show that moraines cover about 186.7 km² of the study area. Rainfall intensity and moraine distribution emerged as the two most influential factors. The simulation gave weights that differ from the standard AHP by less than 1.6%, suggesting the results are stable. The resulting susceptibility map divides the area into four zones: non-susceptible, low, moderate, and high susceptibility. The high-susceptibility zone covers 1656.48 km² (14.26% of the total area), mainly in high-altitude, steep terrain where moraines are present. The model performed reasonably well in validation, with an ROC curve AUC of 0.807. These findings could support local hazard prevention and land-use planning.