In order to design vehicles with diminished gCO2/km emissions level, car manufacturers aim at reducing the weight of their vehicles. One of the solutions advocated by the automotive industry consists in the replacement of metallic parts by lighter systems made of polymer reinforced composites. Unfortunately, the numerical simulations set to evaluate the vibratory and acoustic performances of systems made of this kind of materials are often not sufficiently effective and robust so that convincing test/simulation correlations are rarely met. Indeed, for polymer-based materials, numerous parameters affect the vibroacoustic behavior. On the one hand, it is well known that the viscoelastic properties (Storage -Young- and dissipative moduli) of polymers depend on the temperature, loading frequency and sometimes the humidity content. On the other hand, when focusing of short-fiber composites, the injection molding process leads to an inhomogeneous spatial distribution (density and orientation) of the reinforcing fibers. For instance, through-thickness heterogeneity (orientation and volume fraction) is largely reported. More precisely, near the mid-surface, the volume fraction of fibers is increased and they are mostly oriented perpendicularly to the main direction of the flow. All in all, the composite material is anisotropic and its mechanical properties depend on the geometry in the part and the location within the part.In an industrial context, it is of great importance to rank the influence of the parameters and to set which ones are mandatory in a numerical simulation and which ones are of second order. Thus, the present paper aims at finding guidelines for modeling such complex materials, mainly focusing on the effect of through-thickness reinforcement heterogeneity. The results are based on experimental measurements and numerical simulations performed on an oil pan (typical of car industry) and rectangular plates made of short fiber reinforced polymer composite (SFRPC).