Attenuation models, with and without frequency dependence, have been developed through analysis of time-domain amplitude measurements and teleseismic spectral shape data from Pahute Mesa nuclear explosions. The time-domain analysis is based on a near-field to far-field amplitude comparison. The near-field amplitude information is incorporated in two parameterized explosion source models (Mueller-Murphy and Helmberger-Hadley) based on analyses of near-field data. The teleseismic amplitude observations are from a large data set of WWSSN short-period analog recordings. For the narrow-band time-domain data, the various source and attenuation models are indistinguishable. We utilize the spectral shape data in the 0.5- to 4-Hz band as a constraint on the source-attenuation models at higher frequencies, concluding that either source model, when convolved with the appropriate frequency-dependent Q model, can be consistent with both the near-field and far-field time-domain amplitudes and the spectral shape data. Given the trade-off between source and attenuation models and the similarity of the different source models in the 0.5- to 4-Hz band, it is difficult to prefer clearly one source model over the other. The Mueller-Murphy model is more consistent with surface wave amplitude measurements because of larger predicted long-period energy levels. Whether or not frequency dependence is included in the attenuation model, the value of t* near 1 Hz is about 1.0 sec (assuming the Mueller-Murphy source model) or 0.8 sec (assuming the Helmberger-Hadley source model). This 0.2 sec difference results from greater 1-Hz energy levels for the Mueller-Murphy source model. Adopting an average attenuation model, predicted amplitudes and yields are shown to be within the uncertainty of the data for all the events analyzed.