Production-induced depletion of hydrocarbon reservoirs leads to deformation, compaction, displacements, and stress change in the reservoir as well as in the surrounding rock. Such stress changes affect the acoustic-wave velocity and bulk density. This has two implications. It changes the contrast in acoustic impedance between reservoir and overburden, resulting in seismic amplitude changes at the top of the compacting reservoir. Secondly, it changes the traveltime of seismic reflection waves, leading to arrival-time delays (time shifts) of seismic data gathered in the repeat survey compared to data gathered in the base survey (Hatchell et al, 2004; Kenter et al, 2004; Stammeijer et al, 2004). Maps of time shifts can then indicate the areal distribution of reservoir compaction, and thus reveal the areal distribution of depletion. This could help to locate bypassed oil in undrained compartments, identify drilling targets and sidetracks, and avoid expensive infill wells. These interesting geophysical applications of reservoir mechanics justify questions about how accurate such geomechanical models really are. What is their sensitivity to the (natural) variation in input parameters like geologic structure and sedimentological detail, depletion pattern, and mechanical property distribution? This question is closely related to our ability to capture this variation in computer models via upscaling. Analytical models based on simplified reservoir shapes and linear elasticity show that the reservoir-compaction-induced stress change in the overburden is governed by the contrast in mechanical properties of the reservoir and the rock surrounding it, and by the reservoir shape and size with respect to its depth of burial (Geertsma, 1973; Segall, 1992). Memory and computational capacity of computers now allow numerical analysis with fine-scale (tens to hundreds of meters) geologic reality in geomechanical models.