Aquatic organisms' tolerance to water pollution is widely used to monitor and assess freshwater ecosystem health. Tolerance values (TVs) estimated based on statistical analyses of species-environment relationships are more objective than those assigned by expert opinion. Region-specific TVs are the basis for developing accurate bioassessment metrics particularly in developing countries, where both aquatic biota and their responses to human disturbances have been poorly documented. We used principal component analysis to derive a synthetic gradient for four stressor variables (total nitrogen, total phosphorus, dissolved oxygen, and % silt) based on 286 sampling sites in the Taihu Lake and Qiantang River basins (Yangtze River Delta), China. We used the scores of taxa on the first principal component (PC1), which explained 49.8% of the variance, to estimate the tolerance values (TV(r)) of 163 macroinvertebrates taxa that were collected from at least 20 sites, 81 of which were not included in the Hilsenhoff TV lists (TV(h)) of 1987. All estimates were scaled into the range of 1-10 as in TV(h). Of all the taxa with different TVs, 46.3% of TV(r) were lower and 52.4% were higher than TV(h). TV(r) were significantly (p < 0.01, Fig. 2), but weakly (r(2) = 0.34), correlated with TVh. Seven biotic metrics based on TVr were more strongly correlated with the main stressors and were more effective at discriminating references sites from impacted sites than those based on TV(h). Our results highlight the importance of developing region-specific TVs for macroinvertebrate-based bioassessment and to facilitate assessment of streams in China, particularly in the Yangtze River Delta.