Abstarct: One of the challenging subjects in every carbonate reservoir study is the characterization of reservoir fluids. Water saturation (Sw) is one of the most significant petrophysical parameters required for reservoir management. Water Saturation can be measured directly from core analyses or can be estimated from well logs evaluation. Use of geophysical techniques to characterize reservoirs is becoming increasingly important in the development and production strategies of oil fields. Resistivity values obtained from magnetotelluric (MT) data is sensitive to the electrical properties of rocks with varying fluids and saturations. Hence, achieving a method for water saturation prediction from the resistivity values obtained the MT data is a challenging task. In this research, first, geologic and petrophysical factors controlling the hydrocarbon saturation in the Asmari carbonate reservoir were analyzed, and then, differential effective medium (DEM) was selected to lixnk the electrical properties of the reservoir with its water saturation. This special case of studied reservoir (Vsh=0) led to a simplified version of model baxsed on the Archie’s law. The method of genetic algorithm was used to calculate the Archie’s coefficients in the exploration well named A. A relatively acceptable regression coefficient (0.62) was obtained between the predicted and experimental water saturation. The formation resistivity values of the exploration well A were classified baxsed on the fractal method. The results showed three different zones baxsed on the type of porosity and texture of the rocks. Then, the genetic algorithm was used for calculating the Archie’s coefficients in each of the zones separately. The results indicated that water saturation was dependent on the differences in the reservoir height (H). The reservoir height (H), as a parameter, was introduced in the Archie’s equation, and thus, we obtained a modified Archie’s equation. Using the genetic algorithm for calculating the coefficients of the modified Archie’s equation showed that the regression coefficients of 0.92 and 0.88 were obtained between the predicted and experimental water saturation in the train and test stages, respectively. Artificial neural networks (ANNs) and adaptive neuro-fuzzy inference system (ANFIS) were also applied on the MT resistivity data to estimate water saturation in well No. 115 near the MT line 8805. The regression coefficients of 0.79 and 0.85 were obtained between the predicted and experimental water saturation using ANNs and ANFIS, respectively. It was also concluded that the ANFIS model in comparison with ANN method had more accurate prediction capability for water saturation prediction using the MT resistivity data. The better performance of the ANFIS than the other intelligent methods is easily justified because it is a combination of FIS and ANN methods, and thus, contains the advantages of both methods.