After large earthquakes at subduction zones, the plate interface continues moving due to mostly frictional afterslip or simply afterslip processes. Below approximately 60km depth, the seismic moment release at the plate interface is quite small indicating that the shear strength is low and stable sliding is the prevailing process. This agrees with the lack of significant interseismic locking at deeper segments (>60km) resulting from the inversion of geodetic data and thus low afterslip can be expected. However, inversion models that employ linear viscoelastic mantle rheology and an elastic crust result in significant afterslip at depths >60km. In this paper, we present a combination of a 3D forward geomechanical model with power-law rheology that simulates postseismic relaxation with dislocation creep processes in the crust and upper mantle and an afterslip inversion. We estimate the cumulative viscoelastic relaxation and the afterslip distribution for the first six years following the 2010 M-w 8.8 Maule earthquake in Chile. The cumulative afterslip distribution is obtained from the inversion of the residual surface displacements between the observed displacements from the continuous GPS (cGPS) and the ones from the forward modelling. We investigate five simulations, four with different dislocation creep parameters for the crust, slab, and upper mantle and one with elastic properties for the crust and slab, and a linear viscoelastic upper mantle for comparison. Our preferred simulation considers a weak crust since it shows the best fit to the cumulative cGPS postseismic displacements, a good fit to the time-series, and, in particular, a good spatial correlation between afterslip and aftershock activity. In this simulation, most of the viscoelastic relaxation occurs in the continental lower crust beneath the volcanic arc due to dislocation creep processes. The resulting afterslip pattern from the inversion is reduced at depths >60km, which correlates to the low cumulative seismic moment that is released from aftershocks at these depths. Furthermore, the cumulative afterslip moment release from this simulation corresponds to 10% of the main shock in six years, which is approximately half of the moment release that results from models with an elastic crust and linear viscosity in the upper mantle. We conclude that an integrated analysis by considering power-rheology with dislocation creep processes in the continental crust and upper mantle along with aftershock activity may be used to constrain location and magnitude postseismic relaxation processes better. (c) 2020 Elsevier B.V. All rights reserved.