Abstract: Purpose/Significance The optimal allocation of emergency resources is the prerequisite and guarantee for timely and effective rescue work after natural disasters. Design/Methodology Considering two different demands of disaster-affected areas and refuges, the location of facilities and inventory for different types of relief under uncertain demand are studied. A scenario-based two-stage stochastic programming model is proposed. Based on the tracks of typhoons that landed in Fujian Province from 1949 to 2018, K-means cluster is performed to construct typhoon disaster scenarios. Findings/Conclusions Numerical experiment results show that the established two-stage emergency resource allocation model is practical and feasible, and the solution method ensures the stability and effectiveness of result.