Seismic studies of the deep mantle suffer from the fact that the probing seismic waves must traverse the highly heterogeneous and poorly resolved shallow structure. One potential way forward is to develop high-resolution models of the crust and upper mantle using other information. Here we describe the construction of a geodynamic a priori model of some aspects of upper mantle seismic velocity heterogeneity. It is based on an equal area tomographic grid and it has been produced at two scales, a 1¿ ¿ 1¿ resolution at the equator, (i.e., each cell has an approximate dimension of 100 km by 100 km), and a 5¿ ¿ 5¿ resolution at the equator. Both have a constant layer thickness of 100 km. Currently, the model accommodates the subducting lithosphere and global variation in continental crustal thickness and age of oceanic lithosphere. The shape of subducting oceanic lithosphere was derived from profiles through seismicity. The shape was combined with estimates of plate velocities and age of subducting lithosphere using an analytic solution of the thermal field to define the slab thermal anomaly. The temperature perturbation was converted to a slowness (1/velocity) perturbation. For oceanic lithosphere, a plate-cooling model was used to convert lithosphere age to slowness perturbation via a temperature perturbation. The variation in the thickness of continental crust, around a global average, formed the third element of the slowness perturbation model. This model has already been applied in a high-resolution mantle tomographic study of lower mantle heterogeneity.