The reactivation of paleo-landslides on large reservoir bank slopes has been observed to cause sudden overall instability during progressive deformation processes, resulting in large-scale landslide-barrier lake disaster chains. This phenomenon has been recognized as a critical and challenging issue in disaster prevention and mitigation in reservoir areas of China. In response to this issue, a typical large reservoir bank slope in the Baihetan Reservoir area was selected as a case study, and a novel dynamic stability evaluation model for landslides was established. In this model, stress field variations induced by progressive deformation, such as settlement and inclination changes, were taken into account. Compared with traditional models, the proposed model was found to better reflect the actual stability conditions of the landslide in the field. According to monitoring data, failure occurred between September 1 and October 1, 2021, and stabilization was reached around October 31. During this period, the calculated safety factor exhibited an evolutionary trend of “basically stable” (1.114) → “marginally unstable” (1.041) → “basically stable” (1.122). The reduction in stability was attributed primarily to the combined effects of settlement and inclination changes, among which settlement deformation was identified as the dominant influencing factor. Therefore, in the stability analysis of reservoir bank slopes, it is suggested that, in addition to reservoir water level fluctuations, stress field variations induced by multi-source progressive deformation, such as vertical displacement and inclination—should also be considered. This result has provided a new mechanical modeling basis for early warning and risk assessment systems for large reservoir bank slopes. It is proposed that, in future engineering applications, multi-source monitoring equipment should be deployed at key locations, and the improved evaluation model should be used to monitor real-time variations in displacement and inclination. Such an approach is expected to offer more accurate mechanical criteria for slope hazard assessment in large reservoir projects such as Baihetan, thus facilitating advanced early warning and intelligent risk control.