Chinese rivers are both highly biodiverse and highly under pressure, hence an urgent need exists to understand ecological drivers and disentangle different scales of stressors to support water management. Our aims were to (1) determine the most influential variables for benthic invertebrate occurrence, (2) compare results related to communities as opposed to metrics and (3) examine the role of spatial scales with relevance to management. Benthic invertebrate sampling was performed at 37 sites in selected tributaries of the middle reaches of the Yangtze, covering an environmental gradient with a focus on organic pollution (stratified sampling design). Ten metrics commonly used in biomonitoring were derived and analysed in parallel to assemblage data. Environmental variables covered 74 parameters from three different spatial scales, namely local, reach and catchment scale. We ran a CCA with each of the three subsets to find out the significant determining/explanatory variables, followed by pCCA and pRDA (for metric data) with these variables with forward selection to determine single variables important for each subset; we further used variation portioning for benthic invertebrate data. A high percentage of overall variability (70 %) of the assemblage structure was explained, with catchment- and local-scale variables being almost equally important. Small-scale variables tended to be more important than large-scale variables for the metric-based approach but not for the assemblage approach. Our results emphasise the need for spatially explicit regional studies in freshwater systems and suggest testing multi-metric assessment approaches to tackle water management and environmental health questions in China.