Eutrophication of water bodies is a worldwide ecological problem caused by excessive amounts of nutrients. This paper develops an accounting method for modeling multiple interacting stocks of nutrients, which applies the dynamic nutrient balance approach to calculate empirical nutrient budgets. The proposed model of multiple interacting stocks draws an explicit connection between nutrient stocks on land and nutrient stocks in the sea, which in turn links estimated marine nutrient stocks with the empirical data of nutrient concentrations. The model is applied to the Baltic Sea, where stocks of nitrogen and phosphorus are calculated for six basins, drawing distinction between surface, middle and deep water layers of the basins, and taking into account spatial interactions of nutrients within the water layers and between the basins. The model is calibrated using empirical data on nutrient loads and concentrations. Finally, we apply the model to forecast future development of nitrogen and phosphorus concentrations under different abatement scenarios.