Taihu Lake basin is one of the most populous regions in China, with high urbanization and economic development. This area has faced serious water environmental problems since the 1980's as the amount of pollutants produced and discharged into the rivers and lakes have increased. Non-point source pollution-mainly caused by land use and land cover change-is considered the most important cause of this contamination. Most earlier work researching the relationship between rivers and the landscapes through which they flow in this area simply focused on the impact of land use on water quality at a single spatial scale, and lacked a consideration of natural factors such as geography and climate conditions. To explore the response of water quality to background characteristics of landscapes at a watershed scale, we examined the correlation of 9 landscape factors characterizing land use, stream density, precipitation, and topography with 5 water chemistry indicators of the rivers in Taihu Lake basin. Using redundancy analysis, correlations were conducted at 9 spatial scales from 0.5 km to 24 km, respectively. The results show an overall poor status for the water quality of the rivers, but with an improving trend. The indicators with concentrations exceeding the national standard include total phosphorus (TP), ammonia nitrogen (AN), biochemical oxygen demand (BOD), chemical oxygen demand (COD), and dissolved oxygen (DO). Water quality also varies significantly by location: in the upstream basin, apparent differences are observed between the plains areas with dense stream networks and the forested areas, while in the downstream basin the rivers flowing out of Taihu Lake show a significant difference between their upper and lower reaches. The investigated landscape factors have a significant effect on water quality of the rivers; the water chemistry indicators show scale- and location-dependent responses to the landscape factors. Moreover, different indicators have distinct responses. Specifically, AN, TP, and DO are positively correlated with settlement cover in the upstream basin, but with farmland cover and stream density in the downstream basin. BOD and COD exhibit a negative correlation with natural wetlands and a positive correlation with artificial wetlands in the upstream basin, while being negatively correlated with slope and positively correlated with stream density in the downstream basin. The explanatory power on the variance in water quality of each landscape factor shows apparent scale-dependent features. With increasing spatial scale, the curves of total explanatory power show a consistent pattern in the upstream and downstream basins, presenting double peaks near the scales of 0.5-1 km and 16 km. In the upstream basin, natural wetlands and slope present the highest exploratory power at the smaller and larger scales, respectively. In the downstream basin, stream density has the largest effect at small scales and farmland has the largest effect at a large scales.