The purpose of this research is to establish the landslide risk analysis procedures by which the disaster risk distribution within the watershed was analyzed. Orderly, this research analyzed the landslide hazard, the vulnerability of elements at risk, and the resilience capacity of village to calculate the values of property risk, life risk and total risk possibly caused landslide disasters. First, the landslide hazard assessment included the landslide spatial probability, temporal probability and probability of landslide area. Eleven geomorphological factors and two rainfall factors, evaluated as effective factors because of the higher correlation with the landslide distribution, were regarded in the landslide susceptibility model. The probabilities of landslide area were developed by transforming the power law formula in the landslide frequency-area distribution. Then, the landslide spatial probability and exceedance probability of different rainfall events as well as the probability of landslide area were used to predict annual probability of each slope-unit with a landslide area more than a threshold. Second, vulnerability analysis included the vulnerability of property above land and the vulnerability of human life. The land-use layer was divided into building, house, farmland, forestland, road, water area and no-direct-loss, nine types of element at risk, and the value of its unit area was given against different elements at risk. Then, damage factor was used to represent the average proportion between the actual disaster loss of different elements at risk and the values of elements at risk themselves, and by which property vulnerability was analyzed. The life vulnerability was carried out quantitative analysis by assessment life value and the mortality rate of people in buildings. Third, from the framework, the village resilience capacity consisted of four parts: “the participation experience of disaster prevention drill”, “real-time monitoring system”, “the observation of the one’s own rainfall gauge” and “professional debris-flow volunteer”. Therefore, the scores got from village’s checklist, as well as the weight of every item were used to calculate village resilience capacity. Finally, according to the annual landslide probability and vulnerability result of property, the property risk map was drawn; next the annual landslide probability, vulnerability of life, and resilience capacity were combined to assess life risk; then the total risk map was drawn. This research also carried out risk analysis before and after risk reduction measures held in Shihmen reservoir watershed to get benefit value of different measures. Then benefit-cost analysis was conducted based on benefit value and the cost of risk reduction measures. Villages with high benefit-cost ratios of measures are Hua-Ling, Yi-Cheng, Xiu-Luan and Gao-Yi, which also suffer high annual life risk. The Yi-Cheng village with a high benefit-cost ratio of 192.19 and a low resilience capacity of 0.5716 could reduce its life risk by strengthening the risk reduction measures. The benefit-cost ratio of measures for whole watershed is markedly greater than 1.0. Analytical results showed that the risk reduction measures are cost-effective. Based on the results of sensitivity analysis, the change of property vulnerability and mortality rate would increase the uncertainty of risk values. Additionally, the higher interest rate or shorter life term would decrease the benefit-cost ratio. Nevertheless, the replacement of the mentioned factors does not reverse the cost-effective inference.