Abstract: The central region of Shanxi Province, located in the semi-arid Loess Plateau of northern China, is characterized by severe spatial and temporal disparities in freshwater resource distribution. Rapid urbanization, intensive agricultural activities, and climate variability have exacerbated water scarcity, leading to overexploitation and degradation of aquatic ecosystems. To systematically evaluate the ecological health of river systems in this region, this study conducted a comprehensive survey of macroinvertebrate communities across three major river basins: the Yellow River Basin, Fenhe River Basin, and Hai River Basins in October 2023. A total of 22 sampling sites were selected to represent diverse hydrological conditions, including headwater streams, mid-reach channels, and downstream zones influenced by human's impact. Standardized kick and sweep hand net sampling was conducted using a Sobol net (mesh size: 500 μm) to collect samples of large invertebrates, which were then identified to the lowest possible taxonomic level. Community structure and biodiversity were assessed using seven biological indices: the Shannon-Wiener Diversity Index (H'), Simpson's Dominance Index (d), Margalef's Richness Index (d_m), Pielou's Evenness Index (J), the Biotic Pollution Index (BPI), the EPT Index (Ephemeroptera, Plecoptera, Trichoptera), and the Percentage of Dominant Taxa. Water quality was classified according to China’s national standards (GB 3838-2002), where Grade I represents clean and Grade V indicates severe pollution. Environmental variables, including dissolved oxygen (DO), chemical oxygen demand (COD), total nitrogen (TN), total phosphorus (TP), pH, and watershed-scale factors, were measured to identify drivers of ecological degradation through Redundancy Analysis (RDA). Key findings revealed significant spatial heterogeneity in water quality and biotic responses. A total of 128 macroinvertebrate species from 48 families were identified, dominated by pollution-tolerant taxa (79.69%). The most abundant species included Paraleptophlebia sp. (Ephemeroptera), Caenis sp. (Ephemeroptera), and Cryptochironomus rostratus (Chironomidae), which collectively accounted for 53.4% of total abundance. Sensitive EPT taxa, such as Baetis sp. and Hydropsyche sp., were largely absent from urbanized and agricultural zones. 38.9% of sites were classified as Grade IV (moderate pollution), with elevated COD and TN linked to industrial discharges in Yellow River Basin. 65.6% of sites fell into Grades III–IV (light to moderate pollution) in Fen River Basin, correlating with high agricultural runoff. Upstream regions exhibited the best water quality in Hai River Basin, with only 7.5% of sites categorized as Grade V (heavily polluted), primarily near coal mining areas. RDA ordination explained 68.3% of cumulative variance, with the first axis (39.8%) strongly associated with organic pollution. Agricultural non-point source pollution and population density were key anthropogenic stressors. These results underscore the critical role of organic pollutants and land-use practices in shaping aquatic ecosystem health. To address these challenges, we propose a dual strategy: Immediate Mitigation: Prioritize wastewater treatment upgrades in urban clusters and implement precision agriculture technologies (e.g., controlled-release fertilizers) to reduce nutrient runoff. Long-term Monitoring: establish a basin-wide biomonitoring network using EPT taxa as bioindicators, supplemented by remote sensing for real-time pollution tracking. This study provides a scientific foundation for adaptive management in semi-arid regions, emphasizing the need to reconcile economic development with ecological resilience in the Loess Plateau. Future research should explore seasonal variability in macroinvertebrate assemblages and integrate climate projections to forecast long-term ecosystem trajectories.