A new approach to determine the optimum compound characteristics in each tire material is developed and applied to improve various tire performances following the corresponding objective functions and constraints. In the new approach, optimization technology is combined with a finite element method so that the optimum compound characteristics in each tire material are obtained by minimizing an objective function, while satisfying the constraints. Hence, the targets of rubber characteristics in each material of a tire can be determined by the proposed approach to improve tire performance. When we optimize the compound characteristics in each element of a tire model, the number of design variables becomes more than a few hundred. Therefore, the sensitivity of the stiffness matrix in the optimization is obtained by a semi‐analytical (or implicit differentiation) method and computational time is greatly reduced compared to the finite difference method. The new approach is applied and verified to improve the rolling resistance of a tire. Furthermore, we can obtain guidelines for a new topological tire structure by the Young's modulus optimization, considering that the rubber material should be removed from the initial tire design, if the optimum Young's modulus is almost zero at the element.