Abstract: The measurement performance of ultrasonic gas meters is, theoretically, independent of the object (i.e. the medium) they measure. However, due to the different acoustic impedances of gas and air, the transmittance of sound waves in the two media is different; due to the difference in shear viscosity and molar mass, the attenuation of sound waves is different. A standard piston-type flow device, with a flow range of (0.016～10) m³/h, the maximum working pressure of 10 kPa and expanded uncertainty of U_rel=0.33% (k=2), was used to test three 1.5-level G4 ultrasonic gas meters. The test data showed that 1# had zero drift, which caused its large and medium flow point error to be −5.61% compared to air, and 3# had a wrong-wave and zero-drift phenomenon, which caused its small flow point to have an error of −15.22% compared to air. Therefore, it is recommended that the relationship between gas and air be a mandatory item in type evaluation.