Observations of seismic anisotropy yield some of the most direct constraints available on both past and present-day deformation in the Earth's mantle. Insight into the character of mantle flow can also be gained from the geodynamical modeling of mantle processes on both global and regional scales. We highlight recent progress toward understanding mantle flow from both observations and modeling and discuss outstanding problems and avenues for progress, particularly in the integration of seismological and geodynamical constraints to understand seismic anisotropy and the deformation that produces it. To first order, the predictions of upper mantle anisotropy made by global mantle circulation models match seismological observations well beneath the ocean basins, but the fit is poorer in regions of greater tectonic complexity, such as beneath continental interiors and within subduction systems. In many regions of the upper mantle, models of anisotropy derived from surface waves are seemingly inconsistent with shear wave splitting observations, which suggests that our understanding of complex anisotropic regions remains incomplete. Observations of anisotropy in the D" layer hold promise for improving our understanding of dynamic processes in the deep Earth but much progress remains to be made in characterizing anisotropic structure and relating it to the geometry of flow, geochemical heterogeneity, or phase transitions. Major outstanding problems related to understanding mantle anisotropy remain, particularly regarding the deformation and evolution of continents, the nature of the asthenosphere, subduction zone geodynamics, and the thermo-chemical state of the lowermost mantle. However, we expect that new seismological deployments and closer integration of observations with geodynamical models will yield rapid progress in these areas.