This work focuses on the development of an innovative design methodology for lightweight wheels of road vehicles. In particular, the activity is carried out for the specific case of a wheel designed for an ultra-efficient vehicle for Shell Eco-marathon competition, with the aim of finding preliminary design solutions. A simplified finite element model of the tire structure is employed for an accurate modelling of the forces acting at the tire/rim interface. The material properties of the tire structure are identified by means of experimental tests. The computed tire/rim force distribution is applied to the rim exploiting a simplified finite element model of the wheel rim. A multi-objective optimization problem is formulated, based on mass and compliance minimization. Several wheel design layouts are investigated, which differ in terms of number of spokes (i.e. 3, 5 and 7), spokes layout (i.e. straight and Y-shape) and spokes cross section (i.e. rectangular, C and I). Geometric quantities related to the cross section dimensions of the spokes and to the rim thickness are optimized. Design constraints related to structural stiffness and elastic stability (both global and local buckling) are taken into account. The developed finite-element based model of the wheel is used to train a set of neural networks to approximate the objective functions and the design constraints to reduce the computational effort. A multi-objective genetic algorithm is adopted to obtain the Pareto-optimal solutions. The implemented method has proved to be a valuable tool to support design engineers in taking critical decisions in the early stages of the design process.