The Gurson–Tvergaard–Needleman (GTN) model includes several parameters of different nature. Some of them have micromechanical roots while others are strictly phenomenological. Hence, a methodology should be developed in order to obtain a robust set of parameters with both numerical and physical meaning. The material parameters of the GTN model are found using an identification methodology based on experimental-numerical results. The method is applied on a DC01 ferritic steel sheet. An extended version of the Gurson model, incorporating plastic anisotropy, mixed isotropic-kinematic hardening, void nucleation, growth, coalescence and shear is used. Material parameters are obtained based on an inverse optimization and a sensitivity analysis on specimens with different geometries. The onset of coalescence, reflected as the loss of load carrying capacity in the force-displacement curve, is correctly predicted by the extended GTN model. The strain distribution, obtained by digital image correlation, is also correctly simulated. Nevertheless, the model does not to predict the increase of strain observed experimentally, because of the decoupling between hardening and the damage variable. In this respect, the article provides experimental evidence of the limitations of a GTN model based on a strain-controlled nucleation law with heuristic extensions of hardening.