We propose in this paper an original method for quantifying the material losses at subduction zones mainly based on vertical movements of the active margins. We first test the method, using isostatic and elastic models, applying it to 15 different well‐constrained convergent margins. The main conclusion is that most of them are presently close to the isostatic equilibrium. Consequently, we consider as a first hypothesis that such was also the case in the past and that erosion or accretion are natural mechanisms restoring the equilibrium in response to a perturbation. Then, we apply this model to compute the geometry of the paleoslab knowing the paleotopography of the margin. The comparison between present and past ocean margin lithospheric wedge geometries provides estimates of the removed material that constituted the initial wedge. Two cases were studied for paleoreconstructions: Japan and Peru. The results show that a considerable amount of material was removed during the Neogene along both margins. A detailed interpretation of the Neogene history of the Japan margin is proposed on the basis of our result