Great subduction zone earthquakes provide unique opportunities to quantitatively constrain the rheological structure of the upper mantle. In this study, we have developed a three-dimensional viscoelastic finite element model to study the effects of rheological heterogeneities in the upper mantle on the viscoelastic postseismic deformation of the 1964 Mw9.2 earthquake. The model has determined viscosities of the mantle wedge and oceanic upper mantle to be 3x10^19 Pa s and 5x10^19 Pa s, respectively. The asthenosphere beneath the oceanic plate is optimized to be 80 km thick with a viscosity of 3x10^18 Pa s. The model well reproduced the first-order pattern of the present GPS velocities in Alaska. Further research of this project includes studies such as the block motion, slow slip events and interactions between megathrust and crustal earthquakes in this region.