Significant ecological effects of run-of-river dams on stream chemistry, benthic algae, phytoplankton, macroinvertebrates, macrophytes and fish have been found. However, the quantitative impacts and intensity of such impacts on aquatic ecosystems are still unclear and there was no assessment system to examine such impacts effectively. We investigated benthic diatom communities of 23 run-of-river dams in a Chinese river during the dry season with the goal of developing a diatom-based index of biotic integrity (D-IBI) that would be sensitive to such impacts. Four metrics were selected from 110 diatom attributes of training-site data set to construct the D-IBI based on: (1) significant differences between reference and impaired sites, (2) high separation power and (3) a low coefficient of variation (CV). We then calculated the total D-IBI scores by summing metrics for each site after transformation by 0–10 scaling system. The D-IBI and its metrics were tested using an independent testing-site data set from a tributary of the study catchment. The test results, calculated using the same criteria and scaling system as the training-site data set, indicated significant differences between reference and impaired sites. The developed D-IBI was powerful in terms of separation powers, % sites correctly classified, box-separation ratios, correlation index (CoI) and Cumulative_R2. Then the ecological conditions of the sampling sites were evaluated based on the total D-IBI scores. Overall, the ecological conditions of references sites were “Good”. D-IBI scores were significantly reduced at impaired sites, where affected by the dams, compared with references sites. The newly developed D-IBI effectively signaled impairment of flow regulations in the Xiangxi River of China, and it could be used in future studies measuring the long-term status of streams and the effectiveness of various remediation measures. However, further testing and assessment of their applicability in other regions impacted by small run-of-river dams are still needed.