The volume of material removed by subduction erosion can be estimated quantitatively if the position of the volcanic arc, the position of the paleotrench axis, and a paleo-depth reference surface are known. Estimates based on these parameters along the Japan and Peru Trenches indicate rates of erosion comparable to well-known rates of accretion. Proposed erosional mechanisms along the plate boundary, where horsts on the lower plate abrade the upper one, appear insufficient to handle the minimum volumes of eroded material. Some mechanisms of tectonic erosion at the base of the trench slope can be observed at colliding seamounts and ridges where structures are large enough to be seismically imaged. Local tectonic erosion of the lower slope of the Japan Trench resulted when seamounts entered the subduction zone, uplifted the slope, and oversteepened it. The oversteepened slope failed, debris slumped into the trench axis, and much of it was then subducted. Where a seamount was subducted, a large re-entrant was left in the slope, which filled rapidly by local accretion of abundant sediment. Subduction of the oblique-trending Nazca Ridge off Peru produced many similar structures. Erosion is dominated by uplift and breakup of the lower slope, with subduction of the debris rather than abrasion under high-stress conditions.

Another form of tectonic erosion occurs along the base of the upper plate. Its magnitude is indicated by massive subsidence along the margin; however, because of deep burial, the structure resulting from basal erosion is rarely imaged in seismic records. The volume of material eroded along the base of the upper plate exceeds that eroded from the front of the lower slope.