Urban areas consist of a dense population and prosperous economic activities. People’s property and living facilities are located heavily in urban areas. A disaster can cause tremendous loss of lives and property damages. Limited scientific technology can be used to predict when they might occur. Under such circumstances, a comprehensive disaster prevention plan is important to reduce potential damages caused by the disaster. Refuge base planning is developed based on a suitability assessment system in which influences of disaster potential are considered. The Gray Correlation Theory and Entropy Weight Method were adopted to rationally assess suitability, and improve current methods of refuge base planning. Using an empirical study of Chia-Yi City in central Taiwan, we found that service functions of refuge bases could serve as a priority of suitability improvement in refuge base planning. Fire was found to be the most salient among negative influences of disaster potential, and so it deserves further attention in future planning.