The deep gas reservoirs in the southern margin of the Junggar Basin have tight and poor physical properties, and the water layer resistivity increases due to the invasion of oil-based mud. The accuracy of fluid property identification using resistivity logging is only 50%, making it necessary to explore fluid property identification methods other than electrical logging. Based on theoretical analysis and experimental measurement, a method for identifying the fluid properties of gas reservoirs using acoustic parameters was established. Firstly, the volume modulus and shear modulus of rocks with different gas saturations were obtained based on the GASSMAN equation, SCA model, and Gassman-Hill equation, and the sensitivity of five parameters I_ΔK, I_K, I_Vp/Vs, I_v, and I_Zp to fluid properties was theoretically calculated. Secondly, experiments were conducted to measure the longitudinal and transverse wave velocities under varying saturation conditions and different pressure and temperature in the strata. The experiments examined the sensitivity of five parameters I_ΔK, I_K, I_Vp/Vs, I_v, I_Zp to the fluids. Finally, the logging identification criteria for gas and water layers were given based on crossplot technology, and the fluid property identification work was completed for 7 wells. The results show that the sensitivity of acoustic parameters to fluid properties, from high to low, is I_ΔK>I_K>I_v>I_Zp>I_Vp/Vs; the criteria for identifying gas and water layers can be I_ΔK < 0.5, I_K < 0.85, and V_P/V_S < 1.8; the fluid property identification results of 10 layers in 7 wells show that the identification accuracy based on acoustic parameters is 90%, which can meet the needs of well testing and layer selection in this gas reservoir.