Surface water quality is a key factor in the optimal allocation of river water resources for agricultural use. Water quality indices are widely applied in determining water suitability, particularly in agriculture. However, due to their deterministic nature, these indices often face challenges in accurately monitoring water quality. Therefore, this study compares the changes in water quality of four rivers Zarrin Gol, Ramian, Galikesh, and Chel-Chai in the Gorganrud watershed using the IRWQIsc index through both fuzzy and non-fuzzy methods. To achieve this, water samples were collected biweekly from November 2022 to May 2023, and parameters including pH, EC, DO, NO₃⁻, NH₄⁺, PO₄, and TH were measured according to APHA standards. After normalizing the values using IRWQIsc curves, a fuzzy model based on the Sugeno inference system with trapezoidal membership functions and twenty 'if–then' rules was implemented in Python to analyze data uncertainty and dispersion more effectively. The results showed that the average output values across stations using the fuzzy method were 80.58 with a standard deviation of 5.38 and a coefficient of variation of 0.0667, whereas the non-fuzzy method yielded an overall average of 77.31, a standard deviation of 8.57, and a coefficient of variation of 0.1107. Thus, the fuzzy model provided higher accuracy and consistency. Furthermore, in 75.5% of the samples, both methods yielded similar quality classifications, while in the remaining 24.5%, the fuzzy model demonstrated greater sensitivity in identifying boundary and critical conditions. Additionally, water quality in more than half of the stations was classified as "good" to "very good." Consequently, considering its ability to reduce uncertainty and its higher sensitivity in identifying marginal water quality zones, the fuzzy logic approach can serve as an effective decision-making tool for water resource managers in related planning activities.