Accurate knowledge of groundwater levels and flow conditions in the vicinity of groundwater-dependent terrestrial ecosystems (GWDTE-s) is required for identifying groundwater dependency and comparing the present situation with the optimal one, as part of the status assessment of groundwaters according to the EU Water Framework Directive. Geostatistical methods (like kriging or cokriging) may result in an unrealistic groundwater level map if only a few measured data are available. In this paper a new, grid-based, deterministic method (GSGW-model) is introduced. The aim of the model is to calculate groundwater depth within the required accuracy from sparse data of monitoring wells. The basic principle of the GSGW-model is that the groundwater table is a smoothed replica of the ground surface. Hence, changes in the groundwater level between two grid points are calculated as a function of the digital elevation model (DEM) and soil properties. The GSGW-model was tested in the Nyírség region (Hungary). Results were compared with those gained by ordinary kriging and cokriging. It has been concluded that kriging overestimates the groundwater level in the low part of the test area, where wetlands are located, while the maps produced by the GSGW-model are a better fit of the real variability, providing more reliable estimates of groundwater depth in GWDTE-s as well.