We estimated the site amplification of regional high‐frequency Lg seismic phases by a reverse two‐station (RTS) method using seismic events ( 4–6) recorded by Earthscope’s Transportable Array from 2010 to 2013. We compare regional site amplification estimates (horizontal and vertical) from the RTS technique with horizontal‐to‐vertical spectral ratio (HVSR) estimates derived from ambient noise and earthquake records. We compare the RTS results with (1) shallow shear‐wave velocity estimates from near‐surface (horizontal/vertical) ratios of the local body‐wave (initial P‐wave) method, and (2) high topography, basins, and sediment thicknesses. Our RTS results show a strong positive correlation between regional site amplification and basins such as the Michigan basin, the Illinois basin, and the Mississippi embayment. In the case of the Illinois and Michigan basins, the higher the frequency, the higher the horizontal and vertical amplification. Waves passing through the Appalachian and Ozark plateaus are deamplified on both vertical and horizontal ground components; however, the variation in amplification with frequency is larger for horizontal motion than vertical motion. In some regions, such as the western edge of the Appalachian basin and southern Illinois basin, vertical amplification decreases with frequency but horizontal amplification is essentially invariant with respect to frequency. Topography and sediment thickness are likely to affect amplification and both factors likely frequency dependent. There is a negative correlation between the RTS‐measured amplification and shallow shear‐wave velocity, whereas HVSR shows a negative correlation only for low frequencies . We conclude that regional ground‐motion amplification is clearly a function of more than one variable. In general, it appears that both regional topography (i.e., long‐wavelength topography) and deeper subsurface seismic structures (basins and sediments) have a large impact on site amplification.