A new procedure is described for estimating network-averaged P-wave spectra from short-period, teleseismic recordings of explosions. This procedure is then applied to a large sample of approximately 1000 digitized Nevada Test Site (NTS) explosion seismograms to simultaneously derive estimates of frequency-dependent station correction factors and network-averaged P-wave spectra for each of the 50 selected explosions. An analysis of the station correction factors is then presented which demonstrates that the Veith-Clawson (1972) B(Δ) curve for time domain mb adequately describes the teleseismic distance dependence of the spectral amplitude data over the entire short-period band extending from 0.50 to 2.25 Hz. A simplified scaling analysis of the network-averaged spectra is described and used to confirm the fact that the P-wave source coupling characteristics below the water table at the Yucca Flat and Pahute Mesa testing areas of NTS are essentially identical. We conclude that this new spectral measure can provide robust means for comparing different explosion testing areas.