We re-examine the use of surface-wave magnitudes to determine the yield of underground nuclear explosions and the associated magnitude-yield scaling relationship. We have calculated surface-wave magnitudes for 190 Nevada Test Site (NTS) shots using regional long-period seismograms from a combined super-network of 55 North American stations. Great effort went towards making the data set comprehensive and diverse in terms of yield, source location and shot medium in order to determine the portability of surface-wave magnitude scales. In particular, we examine Pahute Mesa, Rainier Mesa and Yucca Flat explosions detonated above and below the water table, and which range over three orders of magnitude in yield. By observation we find a low-yield measure threshold of approximately one kiloton (kt) for (assumedly) moderately well-coupled explosions recorded at near-regional (<500 km) stations, which have little microseismic noise. In order to utilize regional surface waves (Δ < 15°) for quantifying sources and for discrimination purposes, we have developed related methods for determining time-domain surface-wave magnitudes and scalar moments from regional Rayleigh waves. Employing regional surface-wave data lowers the effective magnitude threshold. One technique employs synthetic seismograms to establish a relationship between the amplitude of the regional Airy phase, or Rayleigh pulse of the Rayleigh wavetrain and an associated surface-wave magnitude, based on conventional M_S determinations, calculated from synthetic seismograms propagated to 40°. The other method uses synthetic seismograms in a similar fashion, but the relationship used is a more straightforward one between scalar moment and peak Rayleigh wave amplitude. Path corrections are readily implemented to both methods. The inclusion of path corrections decreases the M_S variance by a factor of two and affects the absolute scaling relationship by up to a factor of 0.1 magnitude units. This latter effect is attributed to the particular station network used and the Green's functions used to obtain the 40° M_S values. Using a generic structure for the distance travelled past the actual source-receiver path minimizes the difference between magnitudes determined with and without path corrections. The method gives stable M_S values that correlate well with other magnitude scale values over a range of three orders of magnitude in source yield. Our M_S values scale very similarly to more standard teleseismic M_S values from other studies, although the absolute MS values vary by ±M0.5 magnitude units about ours. Such differences are due in part to the choice of MS formula used. For purposes of future user comparisons, we give conversion values to the previous studies. Our most refined M_S values give the relationship M_S = 1.00 x log_10 (yield) +B, where B is dependent upon source region and shot medium. This yield exponent of unity holds for events of all sizes and is in line with M_S-yield scaling relations found by other studies. When events are grouped with respect to source region, significantly better fits to these individual-site linear-regression curves are obtained compared to the fits obtained using a single, all-inclusive- model. This observation implies that shot-site parameters and source structure can significantly affect surface-wave-magnitude measurements. We present these M_S values primarily to augment the extensive historical analysis of explosion data based on surface-wave magnitudes by using regional data to increase the number of events with surface-wave magnitudes. These magnitudes are consistant with the teleseismically determined magnitudes of larger events. We present our preferred surface-wave moment values in a sequel paper.