Abstarct: Site selection is considered as an important factor which affects the growth and development of the firms. So, spatial decision making knowledge try to find precise and comprehensive methods that can determine optimum locations for firms’ activities. Using Automated Teller Machines (ATMs) is the most sensible aspect of changes in banking industries baxsed on new technology. The rapid growth in the use of ATMs may be caused by the great growing in popularity of them among the customers. Despite of high charges in ATMs purchase, setup and maintenance, banks and financial institutes eagerly try to extend the usage of these machines considering the popularity. Two major reasons that can decrease ATMs success are lack of knowledge and management weakness in layout and site selection of ATMs. Therefore, with respect to effective criteria in ATMs site selection, zoning apt regions can lead bank managers and decision makers to select appropriate sites for ATMs. Effective criteria and conditions that influence on decision making of ATMs location must be defined before the procedure determination and tools deployment. So, in the first step of this research, all effective criteria are extracted using research and field survey. To refine and acquire major criteria, expert knowledge helps the research by filling questionnaires. Then, spatial laxyer of selected criteria are prepared using ArcGIS software package. The approach of this thesis against facility site selection specifically ATM location is baxsed on rule-baxsed classification. Considering the uncertainty and ambiguous in decision making process, the rules are baxsed on fuzzy theory. The execution of this procedure needs an efficient algorithm to discover qualified rules from training set and then applying the rules in a fuzzy inference system. Therefore, this research uses Ant Colony Optimization as a well-known way for rule discovery. In the next step, a fuzzy inference system is designed and implemented to classify the data. The result of proposed method is compared with C4.5 algorithm as a popular method in classification rule discovery. The comparisons are performed baxsed on classification accuracy and optimum site selection. To verify the proposed method, both real and artificial data are used. This thesis is an applied research and the noteworthy matter of that is about using advanced artificial intelligence algorithm for ATM site selection in proposed approach which is done for the first time in Iran. The long-term objective of this thesis contains implementation of an intelligence decision support system for ATMs/bank branches site selection.