Implementation of the Water Framework Directive calls for methodologies and tools to quantify nutrient losses from diffuse sources at a river basin district scale. Here, we examine the possibility of using a statistical model for source apportionment and retention of nutrients in a large transboundary drainage basin (44,000 km2). The model approach uses non-linear regression for simultaneous estimation of e.g. source strength, i.e. export coefficients to surface waters, for the different specified land-use or soil categories and retention coefficients for pollutants in a drainage basin. The model was tested on data from 26 water quality stations with corresponding sub-basin data, i.e., land cover, point sources and atmospheric deposition, from the Estonian part of the Lake Peipsi drainage basin. The model showed that it was statistically possible to derive reliable export coefficients (i.e. unit-area loads) for nitrogen on agricultural land and forests. Moreover, it was shown with simple empirical functions that lake retention was approximately 30-35% for both nitrogen and phosphorus and that the riverine retention was low for both nitrogen and phosphorus (approx. 10%). Results show that the MESAW model is a simple and powerful tool for simultaneous estimation of sources and retention of nutrient loads in a river basin.