Water saturation estimation faced a great challenge in tight gas sandstone reservoirs because of the effect of pore structure and strong heterogeneity to rock resistivity. The classic Archie’s equation cannot be always well used. To quantify the effect of pore structure to rock resistivity in tight gas sandstones, taking the Triassic Xujiahe Formation of northwestern Sichuan Basin as an example, 35 core samples were recovered and applied for resistivity experiments in laboratory under the simulated formation temperature and pressure environment, and 18 of them were simultaneously applied for nuclear magnetic resonance (NMR) and high-pressure mercury injection (HPMI) experimental measurements. Relationships between rock pore structure and resistivity parameters were analyzed. The results clearly illustrated that cementation exponent (m) and saturation exponent (n) were heavily affected by pore structure. Rocks with superior pore structure contained relative higher cementation exponent and lower saturation exponent, and vice versa. Afterwards, we raised a parameter of pore size index, which was defined as the ration of macropore and micropore percentage contents, to reflect rock pore structure, and established a model to calculate optimal saturation exponent from NMR data. Meanwhile, various cementation exponent prediction model was also raised by combining with porosity and irreducible water saturation (Swirr). By combining with calculated cementation exponent and saturation exponent, we optimized the Archie’s equation to predict water saturation in our target tight gas sands. Field examples illustrated that the predicted cementation exponent and saturation exponent matched well with core-derived results. The absolute errors between predicted cementation exponent and saturation exponent with core-derived results were lower than 0.05 and 0.07, separately. By using the optimized Archie’s equation, water saturations were precisely predicted from well logging data in our target tight gas sandstone reservoirs.