RF parameter measuring is always done using an averaging method. Averaging may appear in the software of the post data processing algorithm, in the hardware of the embedded DSP (data signal processing) or in the VBW filter of the spectrum analyzer. Different averaging methodologies will have different effects on the final measurement result, and incorrect averaging may cause several DB data errors, which can result in wasted correlation efforts. This presentation will discuss the averaging methodologies frequently used in RF measurement, their differences and when to use each one, using a real world example of a noise figure measurement on Teradyne’s UltrawaveMX8. Keysight’s spectrum analyzer, N9030, is used as the benchmark, and a direct gain method is implemented for the noise figure. A real-world correlation result will demonstrate how to find excellent noise performance on the MX8, and its suitability for the majority of cases with requirements in the frequency range from 50M to 7.5G Hertz.