The secure operation of Gas Insulated Switchgear (GIS) is crucial for the bulk power systems, since the failure of GIS apparatus would cause severe interruption of power transforming and power supply at substations. Therefore, the monitoring and maintenance schemes of GIS devices have gradually taken the reliability-based indices into account, of which Partial Discharge (PD) levels in online monitoring are of substantial significance because PD serves as a major cause of insulation break-down. Ultra High Frequency (UHF) method is becoming widely adopted as the major inspection approach for PD in GIS. This paper presents the experiment-based results of the frequency characteristics of UHF signals caused by PD in typical defects using entire GIS operating system platforms. Specifically, the results covered the propagation attenuation of the UHF signals with peak to peak value and cumulative energy through typical structures, including basin-type insulators, Lshaped branch, isolating switches, current transformers, etc. The results serve as substantial criteria for optimal band width selection for high frequency detection, and as notable concerns to improve the sensitivity of UHF detection and the efficiency of online monitoring by effectively locating the external sensors for PD UHF detection.