The statistical model MESAW was applied to simultaneously estimate export coefficients and retention of nutrients in four Lithuanian and three Estonian river basins (range 946–8,388 km2). This modelling approach uses non-linear regression to yield the export coefficients for total nitrogen and total phosphorus losses to surface waters, considering different land-use classes and retention of nutrients in the river network. The model was applied to data from 40 water quality monitoring sites and corresponding sub-basin data on land use, point sources and atmospheric deposition. The results showed that the studied river basins had a high nutrient retention capacity: 67–78 and 24–63% of total nitrogen and total phosphorus inputs remained in surface waters. The estimated retention was large in lakes: 27–59% for nitrogen and 11–31% for phosphorus. In-stream retention was apparently much lower, in the range 11–15% for total N and 3–12% for total P. Retention in lakes was lower in Estonia than in Lithuania due to the locations and smaller areas of the lakes in the Estonian basins. In Estonia, the highest relative retention in lakes was 46% for nitrogen and 35% for phosphorus. In-stream retention was also somewhat lower than in Lithuania, possibly because of the lower temperature in Estonian rivers in summer.