Using Geographic Information System (GIS) coup overlap operations to carry out Site Selection is more efficient than traditional methods. How to determine the complicated weight of Location factors in GIS site selection model and how to establish a GIS automatic site selection model are both important issues to discuss. In this article, using new department store site selection in Taipei city as an example, we decided weights of Location factors for establishing the GIS site selection model in order to analyze spatial distribution of selected new department store locations in Taipei city. In this study the Location factors and the rule of quantifying data in the GIS site selection model are based on retail theory of Reilly's Principle of Retail Gravitation, Converse's Breaking-point concept, Huff's law and Carey's model. The essential Location factors we chose for the GIS site selection model are Population, Accessibility Attraction, Developing Potential, and Competition. By choosing and transforming the quantifiable data which corresponds to the Location of the site selection, we established GIS site selection sub-models for each Location factor. After GIS overlap operation, it can he seen that the selected department store sites of the resulting map mare or less correspond to reality in Taipei. There are benefits in improving the efficiency of showing spatial distribution for the selected locations. The study also shows that using GIS operation is appropriate for site selection, and can achieve better results.