Structural restoration is commonly used to assess the deformation of geological structures and to reconstruct past basin geometries. Classically, restoration is formulated as a geometric or mechanical problem driven by geometric boundary conditions to flatten the top surface. This paper investigates the use of boundary conditions in restoration to better approach the actual mechanical processes driving geological deformations. For this, we use a reverse-time Stokes-based method with negative time step advection. To be able to compare the results of the restoration to known states of the model, we apply it to a model based on a laboratory analog experiment. In the study, we first test the behavior of the restoration process with Dirichlet boundary conditions such as those often used in geomechanical restoration schemes. To go further, we then relax these boundary conditions by removing direct constraints on velocity and replace them with more “natural” conditions such as Neumann and free-surface conditions. The horizontality of the free surface can then be measured and used as a restoration criterion instead of an imposed condition. The proposed boundary conditions result in a larger impact of the material properties on the restoration results. We then show that the choice of appropriate effective material properties is, therefore, necessary to restore structural models without kinematic boundary conditions.