The mantle deformation inferred from seismic anisotropy can be used to place fundamental constraints on the deformation of the lithosphere, as well as the more general flow in the upper mantle. I will review research relevant to both of these topics, with special emphasis on shear-wave splitting observations, which provide a vertically-integrated characterization of upper mantle deformation. A particularly fruitful approach to exploiting anisotropic observations is to compare the anisotropy-derived mantle deformation field with the surface deformation field derived from geodesy. Two end-member models, vertically coherent deformation and simple asthenospheric flow, predict distinct relationships between these two fields. For the former, mantle deformation resides in the lithosphere, which deforms as a thin viscous sheet. The surface and mantle strain fields are identical in this case. For the latter, the dominant mantle deformation is the shear flow induced by the relative velocity between the lithosphere and the mantle beneath the asthenosphere. For a non-rigid plate, this mantle velocity(magnitude and direction) can be determined. Two regions, Tibet and tectonic North America, typify these two models. In each region, a large-scale surface deformation field has been successfully modeled using geodetic data, known plate driving forces, and reasonable lateral variations in viscosity (Holt, 2000; Flesch et al, 2000). It is found that Tibet is remarkably consistent with vertically coherent deformation extending to depths as great as 250km, a result that casts doubt on a broad class of 'delamination' models that have been proposed for this region. In contrast, simple asthenospheric flow is dominant for much of tectonic North America. The underlying mantle is found to be stationary in a hotspot frame or possibly flowing in a direction opposite to North American absolute motion. This flow thus exerts a drag, rather than driving force on the plate beneath this region.