'Continental interior' was originally defined as a geographic concept on the example of Central Asia. Later James Dwight Dana used it in a geological context, when he argued that North America was an ideal continent with a low, old, stable interior and higher, younger, more active periphery. This picture was thought satisfactory from the viewpoint of fixist tectonics for more than a century, although it was clear that it did not account for the structure of Eurasia. Neither in Asia nor in Europe cratons and/or areas of gentle deformation outside orogenic belts coincide with the continental interior. Mobilist tectonics, and especially its plate tectonics version, made clear that in a world where continents are constantly united and re-dispersed, such non-coincidence with continental interior, however it may be defined, is exactly what one would expect. If, however, after isolation by rifting and/or transform-fault-displacement, no plate boundary cuts across a continent, the lithosphere beneath it would cool and thicken. If this lithosphere is also made up of high Mg/Mg + Fe residual material left after making basalt, it would be lighter than pristine mantle lithosphere. Such lithosphere would be resistant to subduction and to deformation. It would thus protect the portion of the continent overlying it, giving rise to a craton. Cratonic keels of 300 km or deeper can be generated by shortening a depleted mantle cushion that normally reaches down to some -150 km by similar to 50%. In the Archaean, high geothermal gradients would eclogitise the mafic bottom of a continent more mafic than those that are younger and prevent its upper surface from rising during such a shortening. This would keep the tops of Archaean cratons unmetamorphic or at low grades. If a continent with such a deep keel is left alone for a time period on the order of 1 Ga, continuous cooling will render it very strong. Only very large strike-slip systems can remain active as plate boundaries for a long time within a continental interior. Therefore, if a continental interior can be kept away from such a boundary for about 1 Ga, it inevitably will turn into a craton with low relief, probably a high percentage (areawise) of internal drainage, and high climatic continentality. Late Palaeozoic Gondwana-Land was one such region. Pre-Miocene Africa was perhaps another with many features inherited from Gondwana-Land. If India stops pushing, it is likely that Central Asia will turn into yet another one. Plume-controlled active rifting is the only way to destroy such a consolidated continental interior. It is thus the composition and the thermal state plus the state of stress of the lithosphere underlying a continent that determines its tectonic behaviour and not its crustal structure, nor its geography. Craton formation is largely a lithospheric process, not a crustal one. By the very nature of the process of craton formation, cratons commonly, but not always, originate within continental interiors, but continental interior alone can be assigned no tectonic connotation whatever. (C) 1999 Elsevier Science B.V. All rights reserved.