InSAR is particularly sensitive to vertical displacements, which can be important in distinguishing between mechanisms responsible for the postseismic response to large earthquakes (afterslip, viscoelastic relaxation). We produce maps of the surface displacements resulting from the postseismic response to the 2002 Denali Fault earthquake, using data from the Canadian Radarsat-1 satellite from the periods summer 2003, summer 2004 and summer 2005. A peak-to-trough signal of amplitude 4 cm in the satellite line of sight was observed between summer 2003 and summer 2004. By the period between summer 2004 and summer 2005, the displacement rate had dropped below the threshold required for observation with InSAR over a single year. The InSAR observations show that the principal postseismic relaxation process acted at a depth of ∼50 km, equivalent to the top of the mantle. However, the observations are still incapable of distinguishing between distributed (viscoelastic relaxation) and localized (afterslip) deformation. The imposed coseismic stresses are highest in the lower crust and, assuming a Maxwell rheology, a viscosity ratio of at least 5 between lower crust and upper mantle is required to explain the contrast in behaviour. The lowest misfits are produced by mixed models of viscoelastic relaxation in the mantle and shallow afterslip in the upper crust. Profiles perpendicular to the fault show significant asymmetry, which is consistent with differences in rheological structure across the fault.