Strain accumulation and release at a subduction zone are attributed to stick slip on the main thrust zone and steady aseismic slip on the remainder of the plate interface. This process can be described as a superposition of steady state subduction and a repetitive cycle of slip on the main thrust zone, consisting of steady normal slip at the plate convergence rate plus occasional thrust events that recover the accumulated normal slip. Because steady state subduction does not contribute to the deformation at the free surface, deformation observed there is completely equivalent to that produced by the slip cycle alone. The response to that slip is simply the response of a particular earth model to embedded dislocations. For a purely elastic earth model, the deformation cycle consists of a coseismic offset followed by a linear‐in‐time recovery to the initial value during the interval between earthquakes. For an elastic‐viscoelastic earth model (elastic lithosphere over a viscoelastic asthenosphere), the postearthquake recovery is not linear in time. Records of local uplift as a function of time indicate that the long‐term postseismic recovery is approximately linear, suggesting that elastic earth models are adequate to describe the deformation cycle. However, the deformation predicted for a simple elastic half‐space earth model does not reproduce the deformation observed along the subduction zones in Japan at all well if stick slip is restricted to the main thrust zone. As recognized earlier by Shimazaki, Seno, and Kato, the uplift profiles could be explained if stick slip were postulated to extend along the plate interface beyond the main thrust zone to a depth of perhaps 100 km, but independent evidence suggests that stick slip at such depths is unlikely.