The municipal waterworks in Ånes supplies 380 people. The wells are situated about 60 meters west of Landåselva River in a delta at the north end of Lake Randsfjorden. The waterworks has experienced problems with contamination, resulting from manure deposition in the river as well as flooding events. The suspected largest source of contamination is a cattle herd on the east side of the river. The contamination history suggests there is a connection between the river and the aquifer, and thus also a connection between the activities on the two sides of the river that needs to be considered. This contradicts a survey conducted in 1987, which treated the river as a hydraulic barrier. In accordance with the principles stated by drinking water supply regulations in Norway, it is recommended to always choose drinking water sources that have the best possible natural protection against contamination. This requires a well location based on hydrogeological knowledge which includes defining restrictions regarding activities in the well capture zone. In this study, the hydrodynamics of the area were investigated using a numerical model. The work for setting up the model included defining its geometry, boundary conditions and hydrogeological parameters. The geometry was investigated through geophysical surveys, geological maps and borehole information. Results of the analyses produced a stratigraphy of a coarse sand-gravel layer on top of a silty till layer. Through studying the water balance and topography, boundary conditions were set as surface recharge, groundwater fluxes, constant head and a semi-pervious river boundary. The value ranges of the hydrogeological parameters were defined by analyzing pumping test and grain size distribution data. Hydraulic conductivities were found to be in the range of 15-62 m/day. During the calibration process the properties which have a higher impact on the model results were tuned until reaching an optimal solution that resembles the observations. The model shows that river leakage to the aquifer occurs at most times of the year. The effect was greatly increased with increasing pumping rate and river stage. Approximately 10 % of the extracted water from the currently pumped well originated from the river when a realistic pumping rate of 350 m3/day was used. The residence time from river to production well was 20 days, less for higher pumping rates as an effect of their wider capture zones. Sixty days are normally considered necessary for incapacitating intestinal bacteria in the water. A well location which had no interaction with the river was found further west on the delta. Placing a production well in this area at high elevation, and restricting activities in its capture zone would provide a more protected water source with respect to contamination threats. A sensitivity analysis showed that the model was primarily sensitive to changes in constant head boundaries and the introduction of spatially variable hydraulic conductivity. This points to the importance of defining these more accurately in future studies, in order to minimize uncertainties connected with the model.