Abstract: In recent years, the problem of lake eutrophication becomes more and more serious in China, moreover, there are many lakes in China. Furthermore, the geographical environment and degree of eutrophication of the lakes are various among different lakes. So it is urgent to manage and control lake eutrophication regionally in China. Ecoregion of lake nutrients is the basis and premise of lake nutrients and its eutrophication effect control in China. Taking lake areas of Northeast China as a case of the study, an ecoregion method system for lake nutrients and its eutrophication effect which is based on rough set method is discussed in this paper. Firstly, the index system of lake nutrients ecoregion is built up from three aspects such as physiographical environment, ecological index and human activities. Secondly, the rough set method is employed to calculate weights for each index in the ecoregional index system. In this paper, Chl-a/TP and Chl-a/TN are taken as the decision attributes, while factors such as land use situation, DEM, landform, annual accumulated temperature, annual total precipitation, phosphorus content and nitrogen content in the soil, population density and GDP etc. as the condition attributes of rough set method, to work out weights for each index. Thirdly, combining with weights of each index and its value, the assessment score for lake nutrients region in raster format can be figured out. And then, this kind of assessment score is projected to watersheds in the study area. Finally, dual-constraint cluster algorithm is employed to work out the lake nutrient regions by different assessment scores between watersheds. Results show that: 1) Rough set method can figure out the relative weight of each index in an index system from numerous historical data and the relationship between them. So, the influence on the reliability of the result from subjective factors and incompleteness of sample data can be avoided. According to the definition of rough entropy, it is suitable to calculate weights of index for lake nutrients ecoregion that determine objective weight of each attribute by the importance of attribute. 2) Comparing to conventional spatial clustering algorithm, the dual-constrained spatial clustering algorithm considers not only spatial distances between objects but also the similarities of non-spatial attributes between objects. So, eco-regionalization of lakes by nutrient concentration as revealed by the dual-constrained spatial clustering algorithm ensures spatial continuity and score proximity in the partition results. 3) Using the nutrient level data from this study, with the help of rough set method and dual-constrained spatial clustering algorithm, the northeastern humid and semi-humid temperate region of China can be divided into five lake nutrient ecoregions, such as the northern Da Higgan Mts. Mountain Lakes ecoregion, Xiao Hinggan Mountain and Sanjiang Plain Lakes ecoregion, Changbaishan Mountain Lakes ecoregion, the Songnen Plain Lakes ecoregion, and the east Liaoning hilly, west Liaoning hilly, Liaohe Plain Lakes ecoregion.