The geotechnical parameter of soil mass in colluvial landslide is uncertain. Saturated hydraulic conductivity and soil-water characteristic curve (SWCC) are important parameters for the unsaturated seepage analysis. It is of great significance to carry out the reliability analysis of reservoir bank landslide considering the spatial variability. In this study, the Jvzhoudao slope in the Ganjiang Reservoir Area is investigated as a case. Firstly, according to the limited laboratory test data of saturation-matrix suction, the statistical characteristics of the SWCC properties, such as VGM, VGB, VG, and FX model parameters are calibrated by the Bayesian method, and the applied probabilities of SWCC model combination are determined by the laboratory measurements. Secondly, combining the random field characteristics of the saturated hydraulic conductivity with the random characteristics of the SWCC models, spatial distributions of the unsaturated hydraulic conductivity are generated in the slide body. Finally, for the extreme conditions in the Ganjiang Reservoir Area in May 2021, the proposed method was applied to the case stability analysis. Results indicate that the combined effect of rainstorm and water level change has a significant impact on the safety factor, and that the safety factor of the deterministic analysis is low. Reliability index calculated after considering the spatial variability of the unsaturated permeability coefficient cannot meet the specification requirements, while additional slope engineering reinforcement measures should be taken to guarantee the long-term stability of the slope.