We present global models of strain accumulation in mantle flow to compare the predicted finite-strain ellipsoid (FSE) orientations with observed seismic anisotropy. The geographic focus is on oceanic and young continental regions where we expect our models to agree best with azimuthal anisotropy from surface waves. Finite-strain-derived models and alignment with the largest FSE axes lead to better model fits than the hypothesis of alignment of fast propagation orientation with absolute plate motions. Our modelling approach is simplified in that we are using a linear viscosity for flow and assume a simple relationship between strain and anisotropy. However, results are encouraging and suggest that similar models can be used to assess the validity of assumptions inherent in the modelling of mantle convection and lithospheric deformation. Our results substantiate the hypothesis that seismic anisotropy can be used as an indicator for mantle flow; circulation-derived models can contribute to the establishment of a quantitative functional relationship between the two.