Experiments with a surface processes model of large-scale (1–1000 km) long-term (1–100 m.y.) erosional denudation are used to establish the controls on the evolution of a model escarpment that is related to the rifting of a continent. The model describes changes in topographic form as a result of simultaneous short- and long-range mass transport representing hillslope (diffusive) processes and fluvial transport (advection), respectively. Fluvial entrainment is modeled as a first-order kinetic reaction which reflects the credibility of the substrate, and therefore the fluvial system is not necessarily carrying at capacity. One-dimensional and planform models demonstrate that the principal controls on the evolution of an initially steep model escarpment are (1) antecedent topography/drainage; (2) the timescale (or equivalently a length scale) in the fluvial entrainment reaction; (3) the flexural response of the lithosphere to denudation; and (4) the relative efficiencies of the short- and long-range transport processes. When rainfall and substrate lithology are uniform, a significant amount of discharge draining over the escarpment top causes it to degrade. Only when the top of the model escarpment coincides with a drainage divide can escarpment retreat occur for these conditions. An additional requirement for retreat of a model escarpment without decline is a long reaction time scale for fluvial entrainment. This corresponds to a substrate that is hard to detach by fluvial erosion, and therefore to fluvial erosion that is not transport limited. Continuous backtilting of an escarpment due to flexural isostatic uplift in response to denudational unloading helps maintain the scarp top as a divide. It is essential if the escarpment gradient is to be preserved during retreat in a uniform lithology. Low flexural rigidities promote steep and slowly retreating escarpments. For given rainfall and substrate conditions, the morphology of a retreating model escarpment is determined by the ratio of the short-range diffusive and long-range advective transport efficiencies. A low ratio (which is interpreted to correspond to a relatively arid climate and weathering-limited conditions) promotes steep, sharp-topped escarpments with straight main slopes, and escarpment retreat occurs over a wide range of height scales. A high ratio (interpreted to correspond to a more humid, temperate climate) produces a convex upper slope, and concave lower slope morphology and only major escarpments are predicted to preserve a high scarp gradient. Lithological contrasts in the model produce more complex morphologies and predict the formation of scarps crowned by an erosionally resistant caprock. However, resistant caprocks are not an essential requirement for model scarps to retreat. We conclude that the inferred controls and model behavior are both consistent with the present-day morphology of rifted continental margins and with modern conceptual models of landscape evolution.