Several experimental studies have been conducted to determine the NOx reduction by a series of LNT (lean NOx trap) and SCR (selective catalytic reduction) catalytic bricks. An important goal is to minimize the required precious metal loading in the LNT while keeping the NOx emission below a specified level. We present a mathematical model of this system using hydrogen as the reductant. Simulations are used to determine the influence of the architecture of the LNT-SCR bricks, nonuniform precious metal loading in the LNT bricks, and the cycle time at temperatures in the range of 200-350 degrees C. The simulations lead to the following observations: (a) Low temperature reduction is the limiting step in the optimization of precious group metal (PGM) loading in LNT. (b) The NOx conversion increases as the number of the sequential bricks (with total length fixed) increase and reaches an asymptotic limit. From a practical point of view, there is little incentive in using more than two sequential pairs. (c) Nonuniform precious metal loading of the LNT bricks results in only a minor improvement in the deNO(x) performance. (d) The cyde time has a significant impact on the NOx conversion. In the simulated example, the NOx conversion at low temperatures is increased by about 15-20% by reducing the cycle time by a factor of 2. (e) Even at low temperature operation, diffusional limitations in the washcoat are most likely to be important in the LNT but not in the SCR operation. The NOx conversion and ammonia selectivity are reduced when washcoat diffusion is dominant in the LNT.