Dates on ridges with slow spreading rates (1-3 cm/yr), obtained through detailed studies with Deep-Tow instuments and manned submersibles (French-American Mid-Ocean Undersea Study), or where the active structures are not submarine (Afar triangle) yield a precise picture of the axial region. There is evidence for a thinned lithosphere to be present at the axis with a thickness of 4-5 km. The thermal structure and composition of this solid layer can be estimated from seismic data and thermal and petrologic models. Extensional tectonics prevails in a belt some 10 km wide and is expressed at the surface by normal faults and fissures. In most previous mechanical models the existence of the lithosphere at the axis is neglected, and the rise of viscous asthenospheric material in a narrow vertical cleft beneath the axis is considered the main cause for the steady state presence of an axial valley and the development of normal faults. In contrast with these models we suggest here that these features depend on the rheological behavior of the lithosphere at the axis: the lithosphere is continually thinned by tectonic strain but also thickened by cooling at the base and by volcanism at the surface. In the steady state this process can be viewed as a succession of tectonic 'neckings' in the central active part of the axial valley (10 km) followed by doming of the lithosphere over the whole width of the axial valley (30 km), in response to isostatic disequilibrium. Creep is likely to control the process at depth and because of the high temperature and large strains will be of the steady state type. The application of experimental flow laws for material constituting layer 3 and the uppermost mantle to this problem, where both temperature profile and strain rate can be estimated, allows an order of magnitude of the 'strength' of the lithosphere at a given strain rate to be calculated. With this strength, isostatic recovery in response to vertical shear stresses will occur at a distance of about 8-15 km from the axis. The resulting 'simple shear' strain progressively 'levels off' the mean topographic slope until it becomes horizontal in the rift mountains. Whereas for rifted ridges or slowly accreting plate boundaries the behavior of the lithosphere controls the mechanics of the axial region, where only small discontinuous transient magma chambers exist, we suggest that in the case of nonrifted ridges the behavior of the asthenosphere is more important, with axial crust in isostatic equilibrium over a large continuous permanent magma chamber.