To simulate the nutrient loading of Danish coastal waters, river discharge from unmonitored areas must be assessed. In an attempt to identify the most suitable method, for this purpose eight different versions of the three-parameter monthly water balance model “MWB-3” were tested on data from various Danish catchments for the period 1989-97. The model with the highest R2 and the lowest correlation between the model parameters was chosen for subsequent establishment of regression equations between the model parameters and catchment and climate characteristics. The most important explanatory variables of the regressions are soil properties, potential evapotranspiration, groundwater table depth, percentage of wetland area and catchment slope. Mean R2 was 0.82 for the calibration data subset (62 catchments) and 0.79 for the test data subset (22 catchments). Sensitivity tests indicate that the selected model version is more robust for application in catchments dominated by loamy soils than in catchments dominated by sandy soils. Though the MWB-3 model provides a good basis for determining the water balance of Danish catchments at the regional and national levels, modifications and extensions need to be considered for future studies.