The February 21, 2008 Wells, NV earthquake (M 6) was felt throughout eastern Nevada, southern Idaho, and western Utah. The town of Wells sustained significant damage to unreinforced masonry buildings. The earthquake occurred in a region of low seismic hazard with little seismicity, low geodetic strain rates, and few mapped faults. The peak horizontal ground acceleration predicted by the USGS National Seismic Hazard Maps is about 0.2 g at 2% probability of exceedance in 50 years, with the contributions coming mostly from the Ruby Mountain fault and background seismicity (M5-7.0). The hazard model predicts that the probability of occurrence of an M>6 event within 50 km of Wells is about 15% in 100 years. Although the earthquake was inside the USArray Transportable Array network, the nearest on-scale recordings of ground motions from the mainshock were too distant to estimate accelerations in town. The University of Nevada Reno, the University of Utah, and the U.S. Geological Survey deployed portable instruments to capture the ground motions from aftershocks of this rare normal-faulting event. Shaking from a M 4.7 aftershock recorded on portable instruments at distances less than 10 km exceeded 0.3 g, and sustained accelerations above 0.1 g lasted for about 5 seconds. For a magnitude 5 earthquake at 10 km distance the NGA equations predict median peak ground accelerations about 0.1 g. Ground motions from normal faulting earthquakes are poorly represented in the ground motion prediction equations. We compare portable and Transportable Array ground-motion recordings with prediction equations. Advanced National Seismic System stations in Utah recorded ground motions 250 km from the mainshock of about 2% g. The maximum ground motion recorded in Salt Lake City was in the center of the basin. We analyze the spatial variability of ground motions (rock vs. soil) and the influence of the Salt Lake Basin in modifying the ground motions. We then compare this data with the September 28, 2004 Parkfield aftershocks to contrast the differences between strike-slip and normal ground motions.