Gear design techniques usually results on heavy drives that are contradictory to the tendency of mass reduction on the modern road vehicles applications. This paper proposes the utilization of a genetic algorithm to design and optimize geometric parameters of a transmission stage comprised of two steel gears. Although genetic algorithm is not an optimization technique on a straight sense, its use is increasing in several applications where the design space is multidimensional. In this study, the first fitness function proposed is a weighted average of critical performance factors on gears design such as total drive weight, tooth contact stress, bending stress and total space occupied by the assembly and the second fitness proposed considers the center distance influence and the tooth contact stress and bending stress. The weights of the average are modified in several simulations. The results show the influence of this weights on algorithm performance to find out the optimal transmission design.