Abstract-Stiction, which results from contact between surfaces, is a major failure mode in micro electro-mechanical systems (MEMS). Indeed microscopic structures tend to adhere to each other when their surfaces enter into contact and when the restoring forces are unable to overcome the interfacial forces. Since incidental contacts cannot be completely excluded and since contacts between moving parts can be part of the normal operation of some types of MEMS, stiction prediction is an important consideration when designing micro and nano-devices. In this paper a micro-macro multi-scale approach is developed in order to predict possible stiction. At the lower scale, the unloading adhesive contact-distance curves of two interacting rough surfaces are established based on a previously presented model [L. Wu et al., J. Appl. Phys. 106, 113502, 2009]. In this model, dry conditions are assumed and only the van der Waals forces as adhesion source are accounted for. The resulting unloading adhesive contact-distance curves are dependant on the material and on surface properties, such as, elastic modulus, surface energy and on the rough surfaces topography parameters; the standard deviation of asperity heights and the asperities density. At the higher scale, a finite element analysis is considered to determine the residual cantilever beam configuration due to the adhesive forces once contact happened. Toward this end, the adhesive contact-distance curve computed previously is integrated on the surface of the finite elements as a contact law. Effects of design parameters can then be studied for given material and surface properties.