This work deals with the design of electromagnetic absorber of interest in radar absorbing material (RAM) and microwave shielding systems. This absorber is based on a multilayer structure and the parameters taken into account are: number of layers, absorption maximization, microwave angle of incidence and frequency band. Design and optimization of the absorber are carried out using a genetic algorithm (GA) and are based on actual electrical parameters values. Such stochastic method leads to the best trade off between absorption properties and structure thickness, minimizing both the reflection coefficient (RC) and the global multilayer thickness (Thick). All the (carbon based) dielectric materials adopted in GA have been characterized as a function of frequency beforehand and their characteristic impedances inserted in a database (DB) to be available during design stage. In particular, employed materials consist of epoxy resin reinforced with three different carbon species: micro sized granular graphite, single walled carbon nanotubes (SWCNT) and carbon nanofibers (CNF). The morphology of the multilayer structures is fixed a priori, - i.e., composite material type and its location in the multilayer structure is given - and the stochastic optimization procedure can only decide the best thickness of each layer. Main goals have been achieved and the possibility of realizing a quasi-perfect microwave absorber through graphite and nanomaterials has been demonstrated.