Predictive selectivity modelling is used to study fishing gear selectivity. These models rely on fish morphology parameters and fall-through experiments. Here, we developed a model of the mechanical interaction between a fish and a netting mesh during an escape attempt. In this model, the fish is described with three parameters: the first parameter measures its mechanical ability to penetrate a mesh and the others are body stiffness coefficients accounting for fish body deformation under mesh twine pressure. These parameters were identified from the results of a bench test simulating a fall-through experiment through a single 60-mm Polyethylene (PE) mesh. A rigid circular cone, a rigid elliptical cone and a small number of horse mackerel and haddock, offering a limited length range, were tested. Results were generalised and compared with those obtained on a piece of netting. Within the model's assumptions, it is possible to predict the success of an escape attempt if one knows the three fish parameters and the mesh load in normal and transverse directions. The study showed that escape potential is greater through an actual deformable mesh than through a non-deformable mesh. The potential size of this difference depends on the mesh load and orientation.