Agriculture is an important non-point source of water pollution. Limiting the impact of agriculture on water quality requires more effective use of fertilizers and organic manures. Crop simulation models can be a cost-effective tool to optimize the management of fertilization practices, maximizing crop uptake and minimizing soil leaching. The success of these tools requires reasonable accuracy simulating N processes under a wide range of environmental conditions. The purpose of this work is to examine the current simulation of crop N dynamics in CERES-Maize and test some model improvements. CERES-Maize assumes that critical N concentration required for optimum growth, and minimum N concentration below which no growth occurs, change with plant age. The N status of the plant is defined using the actual N concentration in relation to critical and minimum concentrations. Crop N uptake is calculated by comparing the potential soil N supply with the crop N demand. N uptake is partitioned between shoots and roots. During seed growth, the pool of N available for remobilization is compared with the grain demand for growth, and seed N concentration is calculated. We evaluated an alternative approach separating the current shoot N pool into three components: stem, leaf, and ear N pools. Critical and minimum N concentrations on each pool along the season are externalized in the crop species file. The new model provided reasonable simulations when compared with field measurements.