Abstract: Coseismic landslides represent geological phenomena where slopes, either nearing critical states or already in precarious conditions, undergo premature sliding triggered by seismic activity. On 5 September 2022, a M_S 6.8 earthquake struck Luding County, Garze Tibetan Autonomous Prefecture, Sichuan Province, China, triggering numerous landslides and causing severe casualties and property damage. Drawing from field investigations and remote sensing analyses, this paper delves into the interpretation of coseismic landslides within the Luding earthquake's epicentral area. The ArcGIS geographic information system was employed to analyze the spatial distribution pattern of landslides in correlation with diverse factors such as elevation, slope gradient, slope aspect, curvature, lithology, distance to faults, peak ground acceleration, distance to rivers, and distance to roads. To enhance the precision of the Analytic Hierarchy Process(AHP), we introduced Receiver Operating Characteristic(ROC)curves, effectively mitigating the inherent subjectivity in AHP′s determination of controlling factor weights through expert scoring. An improved AHP method was thus employed to assess the susceptibility of coseismic landslides in Luding. The results reveal that the Luding earthquake triggered a total of 5, 008 landslides within the extreme seismic zone, covering an area of 17.28 km², predominantly situated along the slopes of the Dadu River valley. Notably, elevation, distance to rivers, and distance to roads emerged as the pivotal factors dictating the spatial distribution of coseismic landslides. Approximately 26% of the study area was identified as extremely high or high susceptibility zones for coseismic landslides. The reliability of the proposed model was evaluated by calculating the Area Under the ROC Curve(AUC), yielding a value of 0.8505, indicating the model's robustness. This study contributes valuable insights for refining methods to assess the susceptibility of earthquake-induced landslides and provides theoretical underpinning for post-disaster reconstruction planning in Luding.