In nature, mountains can grow and remain as localized tectonic features over long periods of time (>10 m.y.). By contrast, according to current knowledge of lithospheric rheology and neglecting surface processes, any intracontinental range with a width that exceeds that which can be supported by the strength of the lithosphere should collapse within a few tens of millions of years. For example, assuming a quartz-dominated crustal rheology, the relief of a range initially 3 km high and 300--400 km wide is reduced by half in about 15 m.y. as a result of lateral spreading of its crustal root. We suggest that surface processes might actually prevent such a ''subsurface collapse.'' Removal of material from topographic heights and deposition in the foreland oppose spreading of the crustal root and could eventually drive a net influx of material toward the orogeny. We performed a set of numerical experiments in order to validate this hypothesis. A section of a lithosphere, with a brittle-elasto-ductile rheology, initially loaded by a mountain range is submitted to horizontal shortening and to surface processes. If erosion is intense, material is removed more rapidly than it can be supplied by crustal thickening below the range, and the topography is rapidly smoothed. For example, a feature 3 km high and 300--400 km wide is halved in height in about 15 m.y. for an erosion coefficient k=103 m2/yr (the erosion rate is of the order of a few 0.1 mm/yr). This regime might be called ''erosional collapse.'' If erosion is not active enough, the crustal root spreads out laterally and ''subsurface collapse'' occurs. In the third intermediate regime, removal of the material by erosion is dynamically compensated by isostatic rebound and inward flow in the lower crust so that the range can grow. In this ''mountain growth'' regime the range evolves toward a characteristic graded shape that primarily depends on the erosion law. The erosion rate may be high (e.g., 0.5--0.9 mm/yr), close to the rate of tectonic uplift (e.g., 0.7--1.1 mm/yr), and few times higher than the rate of topographic uplift (0.15-- 0.2 mm/yr). These experiments show that surface processes can favor localized crustal shortening and participate in the development of an intracontinental mountain. Surface processes must therefore be taken into account in the interpretation and modeling of long-term deformation of continental lithosphere. Conversely, the mechanical response of the lithosphere must be accounted for when large-scale topographic features are interpreted and modeled in terms of geomorphologic processes. ¿ American Geophysical Union 1996