A simplified semianalytic theory is presented that describes the process of lithosphere thinning through thermomechanical erosion by a hot mantle upwelling or plume. The theory compares well with published numerical models but predicts that the rate of thinning is very dependent on the lateral scale of the upwelling and on the viscosity through the lithosphere-asthenosphere transition. A plume tail is likely to cause rapid local thinning, reducing the lithosphere thickness substantially within a few million years. Plume heads or broadly spreading plume tail material are unlikely to cause significant thinning by this mechanism alone. These results are consistent with geophysical evidence for localized thinning under the Hawaiian volcanic chain but not under the broader Hawaiian swell. They are also qualitatively consistent with the distribution of basaltic volcanism in oceanic and continental hotspot tracks. The theory predicts that thin continental lithosphere may be further thinned sufficiently by a plume tail to cause crustal melting within about 20 m.y. Thus this mechanism may account for some delayed silicic magmatism on some slow continental plates. Higher thinning rates would have occurred in the Archean. ¿ American Geophysical Union 1994