Mantle convection drives plate tectonics and the size, number and thermotectonic age of plates codetermines the convection pattern. However, the degree of coupling of surface deformation and mantle flow is unclear. The use of SKS waves seismic anisotropy shows a coherence of mantle and surface deformation, but significant examples depart from this scenario. We review geological observations and present kinematic reconstructions of the Aegean and Middle East and 3D numerical models to discuss the role of asthenospheric flow in crustal deformation. At the scale of the Mediterranean backarcs, lithosphere-mantle coupling is effective below the most extended regions as shown by the alignment of SKS fast orientations and stretching directions in MCCs. In the Aegean, the directions of mantle flow, crustal stretching and GPS velocities are almost parallel, while, below the main part of the Anatolian plate, SKS fast orientations are oblique to GPS velocities. When considering the long-term geological history of the Tethyan convergent, one can conclude that asthenospheric flow has been an important player. The case of Himalaya and Tibet strongly supports a major contribution of a northward asthenospheric push, with no persistent slab that could drive India after collision, large thrust planes being then decoupling zones between deep convection and surface tectonics. The African plate repeatedly fragmented during its migration, with rifting of large pieces of continents that had then been moving northward faster than Africa (Apulia, Arabia). This also suggests a dominant role of an underlying flow at large scale, dragging and mechanically eroding plates and breaking them into fragments, then passively carried. Mantle flow thus seems to be able to carry plates toward subduction zones, break-away pieces of plates, and deform backarc upper crust where the lithosphere is the thinnest. Most numerical models of lithospheric deformation are designed such that strain is a consequence of kinematic boundary conditions (push or pull on lateral sides), and rarely account for basal stresses due to mantle flow. On the other hand, convection models often treat the lithosphere as a single-layer stagnant lid with vertically undeformable surface. There is thus a gap between convection models and lithospheric-scale geodynamic models. We test different degrees of coupling using 3-D lithospheric deformation models. Preliminary results suggest that lithosphere can be carried by asthenospheric flow, which may lead to plate fragmentation, especially if this flow is applied on a large surface and involves mantle upwelling. However, the presence of a ductile lower crust inhibits the upward transmission of stresses. A highly extended crust in a hot environment such as a backarc domain, with no lithospheric mantle and a ductile lower crust in direct contact with asthenosphere, is more prone to follow the mantle flow than a thick and stratified lithosphere.