The Colorado Plateau is blanketed by Phanerozoic marine and nonmarine strata as young as Cretaceous that are now exposed at elevations of about 2 km. Crustal thickening due to magmatism and horizontal crustal shortening was far less than necessary to cause this uplift, which is commonly attributed to the consequences of mantle lithosphere thinning and heating. The Colorado Plateau and the midcontinent region around Iowa consist of Precambrian bedrock overlain by a similar amount of Paleozoic platformal strata, and thus both regions once had similar lithospheric buoyancy. Mesozoic sedimentation increased the crustal thickness and lithospheric buoyancy of the Colorado Plateau relative to the midcontinent region. Backstripping calculations yield elevation without these sediments and lead to a calculated elevation difference between the two areas of about 1200 m, which represents unexplained plateau uplift. Review of constraints on uplift timing finds little support for a late Cenozoic uplift age and allows early to middle Cenozoic uplift, which is consistent with uplift mechanisms related to low-angle subduction that ended in the middle Cenozoic. Finite element heat flow calculations of low-angle subduction and lithosphere attenuation, using a range of initial lithosphere thicknesses and degree of attenuation, indicate that required uplift can result from tectonic removal of about 120 km of mantle lithosphere from an initially 200-km-thick lithosphere. This allows for partial preservation of North American mantle lithosphere with its distinctive isotopic signature in some late Cenozoic volcanic rocks and is consistent with normal Pn velocities in the uppermost mantle beneath the plateau.