Rural health houses constitute a major provider of some primary health services in the villages of Iran. Given the challenges of providing health services in rural areas, health houses should be established based on the criteria of health network systems (HNSs). The value of these criteria and their precedence over others have not yet been thoroughly investigated. The present study was conducted to propose a model for improving the distribution of rural health houses in HNSs. The present applied study was conducted in Khuzestan province in the southwest of Iran in 2014-2016. First, the descriptive and spatial data required were collected and entered into ArcGIS after modifications, and the Geodatabase was then created. Based on the criteria of the HNS and according to experts' opinions, the main criteria and the sub-criteria for an optimal site selection were determined. To determine the criteria's coefficient of importance (ie, their weight), the main criteria and the sub-criteria were compared in pairs according to experts' opinions. The results of the pairwise comparisons were entered into Expert Choice and the weight of the main criteria and the sub-criteria were determined using the analytic hierarchy process (AHP). The application layers were then formed in geographic information system (GIS). A model was ultimately proposed in the GIS for the optimal distribution of rural health houses by overlaying the weighting layers and the other layers related to villages and rural health houses. Based on the experts' opinions, six criteria were determined as the main criteria for an optimal site selection for rural health houses, including welfare infrastructures, population, dispersion, accessibility, corresponding routes, distance to the rural health center and the absence of natural barriers to accessibility. Of the main criteria proposed, the highest weight was given to "population" (0.506). The priorities suggested in the proposed model for establishing rural health houses are presented within five zoning levels -from excellent to very poor. The results of the study showed that the proposed model can help provide a better picture of the distribution of rural health houses. The GIS is recommended to be used as a means of making the HNS more efficient.