This paper proposes a stochastic linear mixed-integer programming model for flood evacuation planning to optimize decision related to shelter site selection under a hierarchical evacuation concept and probabilistic scenarios. The proposed model considers two criteria as objective function: travel distance and risk index of shelter. This paper not only provides flood shelters and distribution of communities but also determines hierarchical evacuation concept, evacuee's behavior, financial constraint, and uncertainty of flood events. Since problem is formulated, the epsilon constraint approach is selected to solve this problem. We validate the mathematical model by generating a base case scenario using numerical data for Chiang Mai, Thailand. The results are proposed and discussed which represent several alternatives to decision makers. This paper will be great significance in helping decision makers consider the spatial aspect of the strategic placement of flood shelters and flood evacuation planning under uncertainty of flood scenarios.