<p>The potential hazard of seismic landslides is notably high within active fault zones, currently, the commonly used Newmark model for seismic landslide risk assessment often predicts cumulative displacement that are lower than the actual displacement, In order to enhance the earthquake landslide risk assessment accuracy, a new LS-D-Newmark (Landslide density Newmark) model, which considers the attenuation of geotechnical mechanical parameters in areas with different historical landslide densities, is proposed to evaluate the potential seismic landslide hazard. The Xianshuihe fault zone in the eastern Tibetan Plateau was selected as an example, a historical landslide database was established based on fault activity, field investigation, multi-source remote sensing and InSAR monitoring. The landslide hazards in the Xianshuihe fault zone are distributed linearly along the fault zone and are more concentrated at the intersection of the faults. The results of potential seismic landslide risk assessment based on LS-D-Newmark model show that its prediction accuracy (<i>AUC</i> value) increased from 0.78 to 0.84, a 7.69% improvement compared to the traditional Newmark model. Using the spatial characteristics of landslides triggered by the 2022 Luding Ms 6.8 earthquake for verification, and it was found that 75.87% of the landslides were located in the extremely high risk areas and high risk areas predicted by the LS-D-Newmark model, which is consistent with the actual distribution of landslides. The proposed LS-D-Newmark model effectively resolves the issue of underestimating displacement predictions, enhancing the accuracy of potential seismic landslide risk assessments, and provides an important reference for major project planning and construction as well as disaster prevention and mitigation in the region.</p>