The amplification of P-wave amplitudes by the receiver crust was modeled for 34 LRSM (Long Range Seismic Measurement) stations by using crustal structures derived from geological and geophysical information in the literature. If two subsets of stations in western and eastern North America (WNA and ENA), respectively, are taken a linear trend between the computed crustal amplification and the magnitude residuals of Booth et al. (1974) can be seen within each set, but there is a separation of about 0.3 magnitude units between the two groups. The most likely explanation of the separation is anelastic attenuation in the mantle under the western United States. About 75 per cent of the variance in the magnitude residuals, for the U.S. stations used, can be correlated with anelasticity and crustal amplification as known at present. The residual variance must be due to uncertainties in the above factors and other causes.