AbstractThe ecoregion is currently widely used as the basic geospatial unit in freshwater biodiversity conservation. The popularly used delineation is usually based on the assumption that attributes of aquatic ecosystems are influenced by landscape‐scale environmental variables. However, few ecoregion delineations attempt to establish the local validity of this assumption prior to delineation, and few studies check the correspondence of the derived ecoregion boundaries with the distributions of attributes of aquatic biota. In this study, we established an approach to overcome these shortcomings. The notable features of the approach are: (1) the delineation variables were filtered through a series of analytical steps to select those that best represented the aquatic community traits, and which avoided redundancy in the data; (2) the method was quantitative and repeatable; and (3) the derived ecoregion boundaries were checked for consistency with the spatial attributes of aquatic biota. The approach was applied in the Taizi River Basin, northeast China. The procedure proposed here filtered out altitude and annual precipitation as the best variables to include in the freshwater ecoregion delineation. Then, using the quantitative ISODATA classification method, the basin was classified into three ecoregions. A test of accuracy indicated that freshwater ecoregions matched well with the spatial distribution pattern of macroinvertebrate community attributes. Statistical analysis showed that natural geographical attributes and river attributes were different in the three ecoregions, and indices representing macroinvertebrate community attributes are significantly different as a whole among the three ecoregions. The case study proved this approach effective on ecoregion delineation.