Hypoid gears are effectively used in cross axis power transmission systems. Design of hypoid gear parameters is complex and dependent on designers' experiences. In this paper, an easy approach to design the parameters of hypoid gear to obtain the minimum of maximum principle normal contact stress and peak to peak transmission error is presented. An improved Particle Swarm Optimization (PSO) and Back Propagation (BP) algorithm is proposed to predict the stress and the transmission error if certain design parameters are given. The predictive accuracy is evaluated by Root Mean Square Error (RMSE) equation. The results show that the predictive accuracy is in reasonable agreement with the values calculated by the software . Based on the prediction model, the optimization model for the design parameters of hypoid gear is established. This paper proposes a method to design a set of hypoid gears with minimum of maximum principle normal stress and peak to peak transmission error.