The main cause of deterioration for the NO, storage-reduction catalyst (NSR catalyst) is sulfur poisoning. On the basis of the thermogravimetric (TG) and FT-IR analyses of aged catalysts, we assumed sulfur poisoning of NSR catalyst to consist of two main factors. One is that sulfur dioxide in the exhaust gas is oxidized on precious metals and reacts with the support, forming aluminum sulfate. Another is that SO, reacts with the NO, storage components such as barium to form barium sulfate. These consequences lead to the concept that sulfur poisoning should be suppressed by the enhancement of sulfur desorption from the support and barium sulfate. Using a mixture of TiO2 and gamma-Al2O3 as the support minimized the amount of SOx deposit on a catalyst after the sulfur poisoning test. As to gamma-Al2O3, sulfur desorbed at a lower temperature from the catalyst with lithium doped gamma-Al2O3 than the other alkaline or alkali-earth doped gamma-Al2O3 after the sulfur poisoning test. It was effective for enhancing the desorption of sulfur from the aged catalyst. To make a uniform catalytic wash-coat thickness on the substrate, a hexagonal cell monolithic substrate was developed. Hydrogen was the most effective gas for enhancing the reduction of barium sulfate in the aged catalyst, and hydrogen generation on catalyst was enhanced by adding Rh/ZrO2 with high steam reforming reactivity. The catalyst developed by combining these technologies was subjected to on-vehicle durability testing simulating 50,000 km of driving with 30 ppm sulfur fuel to verify the improvement in the NOx purification performance of NSR catalyst.