Groundwater contamination by nitrate associated with fertilization practices is a ubiquitous environmental issue, and consequently of world-wide concern. Controlling this contamination requires an ability to measure and predict nitrate loading from unsaturated zone to saturated zone. A field experiment was conducted in an intensively irrigated agricultural area in Dianchi catchment of Kunming, China. Two celery (Apium graveolens) crop sites with different water table depths (Site A is 2.0 m below the soil surface; Site B is 0.5 m below the soil surface) were selected for the experiment. Both of sites were applied fertilizers at two different rates, one the highest traditionally used by farmers in the region (about 4800 kg N/ha per year, HF) and the other three-eighth of the farmer (1800 kg N/ha per year, LF). The results showed that fertilization practices impacted few effects on the balance and dynamic of water in the plant-soil-aeration zone-saturated zone system. However, groundwater table controlled vertical infiltration recharge and evaporation-transpiration rate. The vertical infiltration recharge and the evaporation-transpiration rate were averagely 0.514 and 5.897 mm/d at Site B with a water table depth of 0.5 m below the soil surface, 0.335 and 6.420 mm/d at Site A with a water table depth of 2.0 m below the soil surface, respectively. Nitrate concentrations of soil water near groundwater table under HF subplot were much higher than that under LH subplot. High fertilization rate consequently resulted in great nitrogen (including nitrate, nitrite and ammonium) loadings from aeration zone to groundwater. At Site B, nitrogen loadings were 316.03 and 223.89 kg/ha a under HF and LF, respectively. Nitrate was the dominant nitrogen component entering groundwater. Little ammonium and less nitrite transported into groundwater. Shallow water table made nitrate entering groundwater more easily and consequently determined the NO(3)(-) loading from vadose zone. For the same fertilization rate, nitrate loading to groundwater under Site B were much higher than those under Site A, with 47.11 kg NO(3)-N/ha a under Site A-HF and 311.73 kg NO(3)-N/ha a under Site B-HF. To avert or minimize the potential of groundwater nitrogen contamination in irrigated agricultural areas should determine and minimize the amounts of applied fertilizer by optimizing them to match crop requirements and environmental protection. (c) 2005 Elsevier B.V. All rights reserved.