The optimal design of auto-catalyst needs a good compromise between the pressure drop and flow uniformity in the substrate. One of the effective methods to achieve this goal is to use the concept of radially variable cell density. But this method has not been examined its usefulness in terms of chemical behavior and conversion performance. In this work, two-dimensional performance prediction of catalyst coupled with turbulent reacting flow simulation has been used to evaluated the benefits of this method n the flow uniformity and conversion efficiency. The results showed that two cell combination of 93cpsc and 62 cpsc was the most effective for improved pressure drop and conversion efficiency due to balanced space velocity and efficient usage of geometric surface area of channels. It was also found that large temperature difference between the bricks in case that the edge of the frontal face of brick has too much lower cell density(less than 67% of cell density of the center of the brick). This study has also demonstrated that the present computational results show the better prediction accuracy in terms of CO, HC and NO conversion efficiencies compared to those of conventional 1-D adiabatic model by comparison with experimental results.