We use Global Positioning System (GPS) velocities and kinematic Finite Element models (FE-models) to infer the state of locking between the converging Nazca and South America plates in South–Central Chile (36°S−46°S) and to evaluate its spatial and temporal variability. GPS velocities provide information on earthquake-cycle deformation over the last decade in areas affected by the megathrust events of 1960 (Mw=9.5) and 2010 (Mw=8.8). Our data confirm that a change in surface velocity patterns of these two seismotectonic segments can be related to their different stages in the seismic cycle: Accordingly, the northern (2010) segment was in a final stage of interseismic loading whereas the southern (1960) segment is still in a postseismic stage and undergoes a prolonged viscoelastic mantle relaxation. After correcting the signals for mantle relaxation, the residual GPS velocity pattern suggests that the plate interface accumulates slip deficit in a spatially and presumably temporally variable way towards the next great event. Though some similarity exist between locking and 1960 coseismic slip, extrapolating the current, decadal scale slip deficit accumulation towards the ~300-yr recurrence times of giant events here does neither yield the slip distribution nor the moment magnitude of the 1960 earthquake. This suggests that either the locking pattern is evolving in time (to reconcile a slip deficit distribution similar to the 1960 earthquake) or that some asperities are not persistent over multiple events. The accumulated moment deficit since 1960 suggests that highly locked patches in the 1960 segment are already capable of producing a M~8 event if triggered to fail by stress transfer from the 2010 event.