The late Cenozoic geomorphic evolution of Grand Canyon has been influenced by three primary tectonic and drainage adjustment events. First, 1 km of relief was produced along the Grand Wash–Wheeler Fault system beginning at 16.5 Ma. Second, the ancestral Colorado River became integrated with the lower Colorado River through Grand Canyon between 5.5 and 6 Ma. Third, the Colorado River was influenced by Plio-Quaternary normal faulting along the Hurricane and Toroweap Faults. Despite the relatively firm constraints available on the timing of these events, the geomorphic evolution of Grand Canyon is still not well constrained. For example, was there a deeply incised gorge in western Grand Canyon before Colorado River integration? How did incision rates vary through time and along the evolving river profile? What is the role of isostatic rebound and Plio-Quaternary faulting on the recent incision history of Grand Canyon? In this paper I describe the results of a process-based numerical modeling study designed to address these questions and to determine the plausibility of different proposed models for the erosional history of Grand Canyon. The numerical model I developed integrates the stream-power model for bedrock channel erosion with cliff retreat and the flexural-isostatic response to erosion. Two end-member paleodrainage and integration scenarios are considered. In the first model, I assume no incision in western Grand Canyon prior to 6 Ma. This model is equivalent to a lake-overtopping scenario for Colorado River integration. In this scenario, the model predicts that Colorado River integration at 6 Ma initiated the formation of a large (700 m) knickpoint that migrated headward at a rate of 100 km/Ma, resulting in rapid incision of western Grand Canyon down to the level of the Redwall Limestone from 6 to 4 Ma and incision of eastern Grand and Marble Canyons from 4 to 2 Ma. Widening of Grand Canyon by cliff retreat triggered flexural-isostatic rebound and renewed river incision of up to 350 m in Plio-Quaternary time according to this model. The model also indicates that Plio-Quaternary normal faulting significantly dampened incision rates in western Grand Canyon relative to eastern Grand Canyon. In the second paleodrainage scenario, I assume that a 13,000 km2 paleodrainage crossed the Grand Wash–Wheeler Fault system at 16.5 Ma. The results of this model scenario indicate that relief production along the Grand Wash–Wheeler Fault system could have initiated the formation of a large (700 m) knickpoint that migrated headward at a rate of 15 km/Ma prior to 6 Ma to form a 150-km-long gorge in western Grand Canyon. Following integration at 6 Ma, the results of this model scenario are broadly similar to those of the first model, i.e., rapid incision through Grand and Marble Canyons from 6 to 2 Ma followed by cliff retreat, isostatic rebound, and fault-controlled incision. The results of the second model scenario illustrate that headward erosion of a proto–Grand Canyon could have been sufficient to capture the ancestral Colorado River east of the Shivwitz Plateau.