Abstarct: Identification of geological infrastructures is an important basic step for dam construction. Modeling is used as a geophysical tool to build two and three-dimensional images of geological structures. Seismic Refraction method is used for measuring the changes in spatial petro-physical parameters such as seismic wave velocity and absorption parameter, baxsed on analysis of first arrival seismic signals. These parameters may depend on the rock material, porosity, joints, alteration and other geological factors. This method is baxsed on studying the propagation of seismic waves and their refraction from different laxyers. In Iran, interpretation of the seismic refraction data is usually done by manual interpretations and by using a series of standard tables. With respect to the importance of power plant projects in the country and high costs which are incurred by early stages of complementary studies, obtaining reliable results of seismic refraction data whose interpretation is done by precise modeling is crucial. Computer-baxsed Inverse Modeling not only offers high precision, but also eliminates likely calculation errors of manual interpretation. It also provides analysts with the opportunity to access Ray Tracing Model and First Arrival Seismic Travel Times so as to assess error of modeling and to compare the processed outputs of the model with the collected data. In this paper, the seismic refraction data is collected from a region near Sourk village in Chahar Mahal-e Bakhtiari Province in which Sork Dam is to be constructed. SeisImager software developed by Geometrics Company is utilized to study geological structure of the region by the model so as to gain overall knowledge of underground structures of the region. The software uses PickWin application to extract First Arrivals and Plotrefa application to create dimensional models, calculate velocity sections, and trace rays. In Plotrefa application a combination of reciprocal time method and least square analysis are utilized for the purpose of modeling. baxsed on observations, there is considerable consistency between the modeling and geological results all of the profiles. Besides, the RMS Error of the modeling of 1 to 8 profiles indicates that all the obtained sections have high correlation with the simulated data. Finally, it can be concluded that this method is suitable for modeling of seismic refraction data and the obtained results correspond well with known geological structures of the region.