We have analyzed shear wave splitting recorded by portable and permanent broadband and long-period stations located in the eastern United States. Teleseismic shear waves (SKS, SKKS, and PKS) were used to retrieve the splitting parameters: the orientation of the fast wave polarization plane ϕ and the delay time Δt. In total, 120 seismic events were processed, allowing for more than 600 splitting measurements. Within the Appalachians, stations located in the western (external) part are characterized by Δt≈1 s and ϕ trending N50¿--70 ¿E in the south and central regions and N30¿--40 ¿E in the north, closely following the trend of the orogenic belt in these areas. The transition region between north and central is characterized by Δt≈1-1.3 s and by E-W trending ϕ that are at a high angle to the regional geologic trend. Measurements at two stations located in the eastern (internal) part of the belt indicate very weak anisotropy. The large-scale pattern of anisotropy is not consistent with that predicted for simple asthenospheric flow beneath the plate. Splitting along the southern and eastern margins of the continent is consistent with that expected for Grenvillian deformation, an alternative model of asthenospheric flow around the cratonic keel cannot be ruled out. Within the cratonic core, the correlation between Δt and lithospheric thickness suggests a lithospheric anisotropy. Smaller-length-scale variations also argue for a significant contribution of lithospheric structures. The fabric responsible for shear wave splitting may have formed during tectonic episodes that affected the eastern United States, i.e., the Grenville and Appalachian orogenies and the subsequent rifting of the North Atlantic Ocean. Our observations in the western Appalachians suggest that the anisotropy may be preserved since the Grenvillian orogeny. The absence of detectable splitting in the two stations in the eastern Appalachians is attributed to the igneous intrusions related to the Atlantic rifting. The measurements in the transition between the northern and central southern Appalachians, constitute an intriguing anomaly, whose E-W ϕ have little obvious relation to the regional surface geology. We suggest two possible causes: (1) the local dominance of asthenospheric flow, motivated by the proximity of a pervasive low-velocity anomaly and (2) lithospheric deformation in a transcontinental strike-slip fault zone active during the Appalachian collision.¿ 1997 American Geophysical Union