A review of the available geological, experimental and theoretical evidence shows that the hightemperature creep of a polycrystalline solid such as the mantle is nonlinear and very close to ideal plasticity. Viscous (Newtonian) convection currents cannot therefore be regarded as a realistic geodynamic process. Thermal considerations indicate that the earth has undergone a radial expansion of at least 100 km (but not more than a few hundred km) since the formation of the oldest datable rocks (3.5·109 years). This expansion has been slowing down in time. The structural unit that has been actively expanding is the interior, where heat accumulates, while the outer shell, a few hundred km thick, has been subject to a tensional deviatoric stress field. Consequently, a condition of plastic instability has developed in the outer shell, which has originated elongated tensile features. This process accounts for the formation of some new crust, but not in the amount required by the continental drift hypothesis. Also, plastic convection can take place in the rheosphere (100-400 km of depth), where the ductility is locally increased by thermal and/or chemical factors. However, the convection pattern indicates that this is a local phenomenon unlikely to affect material outside a few mobile belts in the tectonosphere. Continental drift, accordingly, can be brought about only by a decoupling of the lithosphere from its substratum. The simplest way in which horizontal displacements can be accomplished is by the action of the force of gravity (gravity megatectonics). The required slope at the surface of decoupling is very small (less than 1 4 of a degree), and undations at the lithosphere-rheosphere boundary can be originated by a variety of processes. The equality between oceanic and continental heat flow, which is the main obstacle to the continental drift hypothesis, can be reconciled with it if the continents have a thermal blanketing effect on the underlying mantle. Thus, it appears that lithospheric plate drifting is physically possible. This conclusion, however, merely establishes a possibility, and claims that the problems of global tectonics have been "solved" have not yet been substantiated.