Context: Untangling relationships between landscape patterns shaped by human stressors and related response of fish communities is important for identifying biodiversity patterns and conservation targets, yet in large rivers this knowledge is extremely limited. Objectives: Our study focuses on how human stressors within a riparian landscape zone, including both riparian land use and in-channel stressors, explained the fish community structure in a large river. Methods: We studied fish community patterns along the upstream-downstream gradient of the Yangtze River, China. The curve estimation was used to test correlations between fish metrics and the distance from the estuary. We linked human stressors to fish metrics by multivariate generalized linear models. Results: We collected a total of 63 freshwater fish species from 6,147 specimens. Limnophilic species had the highest richness, represented by 30 specie. The predominant riparian land uses in the studied reaches were cropland (65.3% ± 13.1%) and urban land (19.7% ± 13.6%). There were strong negative correlations between riparian land use (e.g., urban land) and in-channel stressors (e.g., shoreline modification, navigation, and fishing pressure) and fish assemblages, especially limnophilic fish abundance, biomass, and richness. Conclusions: These results demonstrate influences of both riparian land use and in-channel stressors on fish communities, and highlight the use of landscape descriptors as a valuable approach to assess linkages between human pressures and fish diversity in large river systems. Management recommendations include: establishing or rehabilitating riparian buffers, improving commercial navigation management, implementing shoreline protection measures, and reinforcing fishing laws and regulations.