Ground-motion attenuation relations have an important impact on seismic hazard analyses. Ground-motion modeling is particularly sensitive to assumptions about wave-propagation attenuation (crustal Q and geometrical spreading), as well as source and site conditions. Studies of path attenuation from earthquakes in eastern North America (ENA) provide insights into the appropriateness of specific attenuation relations. An Electric Power Research Institute (EPRI) (2003, 2004) study combines published ENA ground-motion attenuation relations into four model forms: single-corner, double-corner, hybrid-empirical, and finite-fault. When substituted in the U.S. Geological Survey 2002 national seismic hazard maps for the five ENA relations originally used in those hazard calculations, the EPRI (2003) relations predict similar ground motions and hazard at short periods (<0.5 sec) and significantly lower ground motions (20%-50%) at longer periods (>0.5 sec), relative to the 2002 national maps. A major reason for this difference is due to the crustal seismic-wave attenuation model assumed in a few of the ENA relations combined into the EPRI (2003, 2004) models. Although appropriate differences in geometrical spreading models among ENA relations can also be significant, a few ENA relations have 1-Hz Q-values (Q0) that are below the EPRI (1993) consensus range for Q0 when coupled with a geometrical spreading of R-0.5. The EPRI (2003, 2004) single-corner relation is strongly influenced by the inclusion of ENA relations with assumed Q0 below the EPRI (1993) range, which explains much of the discrepancy in predictions at longer periods.