In this study we investigated the origin of seismic anisotropy in the mantle beneath North America. In particular, we evaluated whether shear wave splitting patterns in eastern North America are better explained by anisotropy caused by lithospheric deformation, anisotropy due to mantle flow beneath the lithosphere, or a combination of both. We examined new measurements of shear wave splitting from the Missouri to Massachusetts broadband seismometer array (MOMA), the North American Mantle Anisotropy and Discontinuity experiment (NOMAD), as well as splitting parameters from several previous studies. We developed a simple finite difference model that approximates mantle flow around a complex, three-dimensional continental lithospheric keel. To evaluate potential anisotropy from mantle flow beneath the lithosphere in eastern North America, we compared shear wave splitting observations to predicted splitting parameters calculated using this mantle flow model. Our results indicate that a significant portion of observed shear wave splitting in eastern North America can be explained by mantle flow around the continental keel. However, shear wave splitting patterns in a few regions of eastern North America indicate that a component of lithospheric anisotropy must exist, particularly in regions containing the largest keel thicknesses. For eastern North America, as well as for splitting observations in Australia, Europe, and South America, we favor a model in which anisotropy is controlled by a combination of both lithospheric deformation and subcontinental mantle flow. ¿ 2000 American Geophysical Union