All water quality indicators have a definite answer for the water quality of the stations. The use of fuzzy inference system has eliminated the limitation of the definiteness of the answers and the lack of consideration of the parameters of water quality determination from the standards. In this paper, data from 1990-89 in 12 stations in the upper and lower range, Mahabad Reservoir Dam, and seven parameters of turbidity, electrical conductivity, chemical oxygen requirement, biochemical required oxygen, total hardness, ammonium and phosphate, Used. All data were received from the Regional Water Authority of West Azerbaijan Province. MATLAB software has been used to perform analyzes. The Mamdani model has been used for water quality classification and triangular membership functions for model inputs and outputs. The input membership functions are three and the output membership functions are five. From the World Health Organization standard and existing standards for the quality of surface water resources in Iran to determine the desired, acceptable and unacceptable range for input parameters as well as the operator and to define the rules of use and with the help of fuzzy inference system of water quality index (FWQI) development given. The SW9 station is the most polluted and the SW6 station is the least polluted. The results show that the fuzzy index is appropriate in estimating surface water quality in the study area.