I review recent progress in using a range of constraints from structural seismology for understanding the Earth's mantle system. Such constraints can come from heterogeneity power spectra, correlation analysis between global tomography and geodynamic forward models, and comparisons of flow model predictions with seismic anisotropy, for example. Such analsis can yield information on the mode of mantle convection, and in particular the study of seismic anisotropy is promising to be a high resolution tool to analyze lithosphere-asthenosphere interactions. It may be possible to construct geodynamic reference models and to probe these models quantitatively via their deviations from the seismological images, learning about mantle rheology, intra-plate deformation and asthenospheric hydration state in that process.