Although gravitational collapse has been widely accepted as a viable explanation for both synorogenic and postorogenic extension, its mechanisms have not been well understood. Previous models of continental deformation that assumed a viscous or viscoplastic rheology are not well suited for simulating stress states and brittle deformation within the upper crust, where most extensional collapse occurs. Here we present preliminary results of a three-dimensional (3-D) finite element modeling of stress and faulting patterns within the orogenic crust using a viscoelastic rheology. Major parameters controlling orogenic extension, including topographic loading, tectonic compression, basal shear, thermal structure, and 3-D tectonic boundary conditions are systematically explored in the model. For typical tectonic compression (60--100 MPa) and >2--3 km elevation the model predicts synorogenic extension in high plateaus and concomitant thrusting in the lowlands near the foothills. Viscous relaxation within the ductile crust may amplify deviatoric stresses within the brittle crust by a few times, and stress amplification is greater when the crust is hotter, and therefore the brittle crust is thinner. Most synorogenic extension occurs in the direction orthogonal to regional compression, while postorogenic extension is more likely to occur normal to the trend of mountain belts. We apply the model results to the Tibetan plateau and suggest that the age of the north trending grabens is not a reliable proxy for the time when the plateau had reached it present mean high elevation. Conversely, development of the east trending South Tibetan Detachment system in the Miocene by gravitational collapse was more difficult and may require a higher elevation of the Himalayas than present or much of the Tibetan plateau to have remained low till late Miocene. ¿ 2000 American Geophysical Union