Many structural designs (laminated structures, bonded joints) involve the presence of weak in- terface zones, where a crack can initiate and propagate catastrophically. Such possibly unstable response constitutes a strong design concern in the development of these structures. In most practical situations, the presence of a pre-existing flaw or stress concentrator leads to an initial structural response which is governed by the energy dissipated during crack propa- gation, as predicted by Griffith’s theory. The design of microscale features leading to interface dissipation mechanisms at different length scales can contribute to delay crack propagation and improve the structural response. In this context, numerical simulations [1] help define the required macroscopic interface prop- erties with respect to the structural application at hand, while experimental tests [2] explore the pertinence and robustness of some practical implementations of this concept in the context of long fibers, polymer matrix composites. The strong dependence of the nonlinear structural response on both local microscale features and structural configuration is highlighted both nu- merically and experimentally, making the bond design a strongly coupled material-structural problem.