[Objective] The factors influencing landslide potential of loess terraces and the associated landslide formation mechanisms were studied in order to provide a scientific basis for the prevention and control of loess terrace landslides. [Methods] With data from the Google Earth Engine cloud computing platform for Pengyang County, Ningxia Hui Autonomous Region, we used the random forest algorithm to extract terrace distribution information in the study area. Based on the overlay analysis method, the high and extremely high susceptibility zones of potential loess terrace landslides in the study area were determined. Eight influencing factors were selected, and the certainty factor method was used for analysis. [Results] ① The terrace area in the study area accounted for 47.28% of the total area of the county. There were 86 potentially unstable slopes, and there were no deforming areas within the influence range of terraces. ② The high and extremely high susceptibility zones for potential landslides of loess terraces accounted for 46.68% of the total area of high and extremely high susceptibility zones in the study area (27.64% of the total area of terraces in the study area). ③ Rainfall, stratum lithology, gully density, and slope direction had strong controlling effects on the spatial distribution of potential landslides of loess terraces in the study area. [Conclusion] The main controlling factors for potential landslides of loess terraces in the study area were rainfall and stratum lithology, especially in the area where the annual rainfall was more than 450 mm and there were Quaternary stone loess with paleosol strata.