Pn travel times are affected not only by lateral variations in crust and mantle velocity but also by significant amounts of laterally varying anisotropy. To investigate uppermost mantle anisotropy, a tomography algorithm was reformulated to include lateral variations in both velocity and horizontal anisotropy, and it was applied to Pn travel time data from the western United States. Results show that anisotropy is as important in explaining the travel time residuals as are the velocity variations. A detailed resolution study examined the trade-off between the velocity variations and the anisotropy variations and showed that both could be resolved for regions with good ray path coverage. Pn anisotropy beneath the western United States has maximum amplitudes of ¿0.3 km/s (7.5%) when resolved on a length scales of around 3¿. The orientations of Pn anisotropy often correlate well with those inferred from shear-wave splitting results. This correlation suggests that orientation of mantle anisotropy does not change significantly with depth for many regions. The orientation of the Pn anisotropy can be correlated with some of the tectonic processes which occur within the western United States. For example, the fast direction of Pn anisotropy parallels the northeast direction of subduction of the Juan de Fuca Plate beneath the northwest Pacific coast, suggesting that there the anisotropy is related to subduction-driven deformation. The fast direction of Pn anisotropy also parallels the strike of the San Andreas Fault system in central California, indicating that shear strains along the plate boundary may be responsible for the anisotropy there. Within the Great Basin, the Pn anisotropy varies substantially, and both partial melting and small-scale convection within the uppermost mantle could be responsible for these anisotropy variations. ¿ American Geophysical Union 1996