ABSTRACTOut-of-plane response of unreinforced masonry elements is frequently the most critical aspect of the seismic performance of existing masonry buildings. The response of such elements is usually governed by equilibrium rather than strength. Hence, it is customary to resort to rigid-body models, accounting for possible rotations, and/or sliding. However, the results of such analyses depend on the initial choice of the mechanism. In this article, the shaking-table experiments on a brick-masonry specimen, and on a stone-masonry specimen have been modeled by resorting to a combined finite-discrete element strategy. Despite the coarse discretization of both discrete and finite elements, the three-dimensional models are able to capture the experimentally observed multi-degree-of-freedom mechanisms, without any a priori assumption on the mechanism. A sensitivity analysis is carried out, addressing eight different parameters. The identification of the mechanism is sufficiently robust, but the assessment of its activation and failure is best done by combining the finite-discrete element model with a simplified model of the recognised mechanism.