A variety of near-regional, regional, and teleseismic ground-motion data have been used to evaluate proposed models of the nuclear seismic source function for underground detonations in tuff/rhyolite emplacement media. It has been found that both the near-regional broad-band seismic data and the teleseismic body-wave magnitude data are consistent with the modified source model proposed by Mueller and Murphy (1971) but not with the simple cube-root of the yield-scaling source model. In particular, the observed linearity and slopes of the body-wave magnitude-yield curves as well as the observed variation of P-wave period with yield have been found to be fully compatible with the modified source model. On the other hand, it has been concluded that the observed long-period surface-wave data are inconsistent with a simple, spherically symmetric source model. The results of a preliminary analysis have suggested that this discrepancy may be related to the spall closure phenomenon.