Quantitative risk assessment of landslides requires the probabilistic evaluation of spatial impact of run-out processes. To that end, it is necessary to estimate landslide run-out intensities (e.g., extension, deposition height, velocities) using probability methods, i.e., probabilistic landslide run-out hazard evaluation. In this paper, a general framework is presented to estimate three-dimensional spatial impact probability for landslide run-out. In the presented framework, a continuum mechanics concept-based dynamic numerical model namely Massflow, is used to conduct run-out analyses. However, the proper selection of model input parameters still remains the biggest challenge in the run-out modeling because they are tied to large uncertainties. To systematically quantify those uncertainties, model input parameters are defined as random variables subject to reasonable probability distributions (i.e., prior distributions), and their distributions are efficiently improved through Bayesian inference with multiple-observations (observed deposition heights at different points in the past landslide accumulation area). Then the calibrated distributions (i.e., posterior distributions) can be used as an input with reliability methods to estimate run-out exceedance probability (i.e., spatial impact probability of landslide run-out) of the potential landslide similar to the past landslide, and to produce run-out hazard maps. Kriging-based surrogate models are used to improve the computational efficiency. To illustrate the performance of the proposed framework, we apply it to two successive sliding of landslide CJ#8, which is located in the Heifangtai, Gansu, China. The computed results are used to produce probabilistic run-out hazard maps of maximum run-out intensities exceeding various run-out height thresholds (i.e., 0 m and 1.5 m) and various run-out velocity thresholds (i.e., 1.5 m/s and 3 m/s). The maps can be visualized in a GIS platform, and superimposed on top of the landslide numerical elevation model or satellite image map, making it easy for decision makers to know the areas that may be affected by the landslide. In addition, the computed results can be used for probabilistic vulnerability evaluation in the quantitative risk assessment. Probabilistic risk analyses allow a cost–benefit analysis-based prioritization of mitigation strategies and data support for land planning and management.