Lithospheric stress magnitudes may be estimated through the application of geological thermobarometers <Mercier, 1980> and grain size piezometers <Ross et al., 1980). Discontinuities in grain size profiles based on series of peridotite xenoliths from some representative localities can hardly correspond to abrupt changes in stress conditions at depth, inasmuch as no geochemical difference is observed across the textural discontinuities, and no phase changes are possible. These discontinuities would thus result from reaching a critical temperature or stress beyond which the dominant recrystallization mechanism drastically changes. Indeed the grain size piezometer is very sensitive to the recrystallization mechanism <Porier and Guillipe, 1979>, and whereas recrystallization through subgrain rotation has been recognized in some natural periodoties Nicolas et al., 1971>, the experimental data available so far apply to grain boundary migration recrystallization. On the basis of two significantly different grain size profile discontinuities, an empirical relationship (&sgr;=74.5/D) is inferred for the grain size resulting from subgrain rotation. Applying this new relationship when relevant, the stress magnitudes estimated from deep samples remain virtually changed relative to previous results <Ave Lallemant etal., 1980>, as do their implications regarding large-scale convection. However, revised magnitudes for the uppermost mantle may be significantly increased to a value as high as 45 MPa (450 bar) for the Basin and Range Province, with stresses in the lower crust possibly reaching up to 80 MPa.