In this study, we have constructed an empirical least-squares model for maximum seismic ground velocity (vertical component), as a function of source size and distance, from explosions in drillholes and bodies of water. This model is intended, in part, for use in designing explosion-seismic experiments in order to produce acceptable signal-to-noise ratios at larger distances (> 10 km) and to avoid damaging human structures at smaller distances (< 1 km). Data consist of velocities recorded by 120 portable seismographs from 159 shots fired by the U.S. Geological Survey in Alaska and Arizona. Results show that maximum ground-velocity amplitude (not necessarily first-arrival amplitude) decreases with distance, and the functional dependence changes with distance; for example, at 3 km, amplitudes decrease as r−2, but at 73 km, they decrease as r−1. Results also show a strong dependence on site conditions at the shotpoint; for example, a drillhole shot in wet alluvium produces amplitudes that are four times larger than a drillhole shot in dry alluvium. Ground velocities are proportional to the square root of the weight of explosives.