Contamination of groundwater resources by nitrate leaching under agricultural land is probably the most troublesome agriculture-related water contamination worldwide. Contaminated areas often show large spatial variability of nitrate concentration in wells. In this study, we tried to assess whether this spatial variability can be characterized on the basis of land use and standard agricultural practices. Deep soil sampling (10 m) was used to calibrate vertical flow and nitrogen-transport numerical models of the unsaturated zone under different agricultural land uses. Vegetable fields (potato and strawberry) and deciduous orchards (persimmon) in the Sharon area overlying the coastal aquifer of Israel were examined. Average nitrate–nitrogen fluxes below vegetable fields were 210–290 kg ha−1 yr−1 and under deciduous orchards were 110–140 kg ha−1 yr−1. The output water and nitrate–nitrogen fluxes of the unsaturated-zone models were used as input data for a three-dimensional flow and nitrate-transport model in the aquifer under an area of 13.3 km2 of agricultural land. The area was subdivided into four agricultural land uses: vegetables, deciduous orchards, citrus orchards, and non-cultivated. Fluxes of water and nitrate–nitrogen below citrus orchards were taken from a previous study in the area. The groundwater flow model was calibrated to well heads by changing the hydraulic conductivity. The nitrate-transport model, which was fed by the above-mentioned models of the unsaturated zone, succeeded in reconstructing the average nitrate concentration in the wells. However, this transport model failed in calculating the high concentrations in the most contaminated wells and the large spatial variability of nitrate concentrations in the aquifer. To reconstruct the spatial variability and enable predictions, nitrate fluxes from the unsaturated zone were multiplied by local multipliers. This action was rationalized by the fact that the high concentrations in some wells cannot be explained by regular agricultural activity and are probably due to malfunctions in the well area. Prediction of the nitrate concentration 40 years in the future with three nitrogen-fertilization scenarios showed that (i) under the business as usual fertilization scenario, the nitrate concentration (as NO3−) will increase on average by 19 mg L−1; (ii) under a scenario of 25 % reduction of nitrogen fertilization, the nitrate concentration in the aquifer will stabilize; (iii) with a 50 % reduction of nitrogen fertilization, the nitrate concentration will decrease on average by 18 mg L−1.