Here, I present a new absolute plate motion model of the Earth's surface, determined from the alignment of present-day surface motions with 474 published shear wave (i.e., SKS) splitting orientations. When limited to oceanic islands and cratons, splitting orientations are assumed to reflect anisotropy in the asthenosphere caused by the differential motion between lithosphere and mesosphere. The best fit model predicts a 0.2065°/Ma counterclockwise net rotation of the lithosphere as a whole, which revolves around a pole at 57.6°S and 63.2°E. This net rotation is particularly well constrained by data on cratons and/or in the Indo-Atlantic region. The average data misfit is 19° and 24° for oceanic and cratonic areas, respectively, but the normalized root-mean-square misfits are about equal at 5.4 and 5.2. Predicted plate motions are very consistent with recent hot spot track azimuths (<8° on many plates), except for the slowest moving plates (Antarctica, Africa, and Eurasia). The difference in hot spot propagation vectors and plate velocities describes the motion of hot spots (i.e., their underlying plumes). For most hot spots that move significantly, the motions are considerably smaller than and antiparallel to the absolute plate velocity. Only when the origin depth of the plume is considered can the hot spot motions be explained in terms of mantle flow. The results are largely consistent with independent evidence of subasthenospheric mantle flow and asthenospheric return flow near spreading ridges. The results suggest that, at least where hot spots are, the lithosphere is decoupled from the mesosphere, including in western North America.