Abstract: Understanding the response of water quality in lake to landscape pattern at different spatial scales in agricultural watershed is of great significance to water quality management. In this study, we classified seven riparian buffer zones of lakeside zone by ArcGIS and RS in the Honghu Lake, according to the five functional areas. The landscape metrics were analyzed at multiple buffer widths using Fragstats software. Mathematical statistical methods and models such as redundancy analysis were used to explore spatial relationship between water quality and landscape patterns. Results showed that: 1) The effect of landscape patterns on water quality was scale-dependent at multiple buffer widths. The highest total explanatory power between landscape characteristics and water quality was found at the 200 m buffer width, accounting for 86.1% of the total, which was the most effective spatial scale affecting water quality. 2) The landscape configuration (e.g., largest patch index, patch density) was more associated with water quality than landscape composition (e.g., the percent of landscape and evenness index). 3) The impacts of different landscape types on water quality varied. Agriculture land, affected by topography and cultivation mode, was the main influencing factor on the degradation of water quality at smaller buffer widths from 100 m to 500 m. Forests with higher density and area had more purification effect on water pollutants at the wider buffer widths from 1000 m to 5000 m. The impacts of grassland on water quality was similar with that of forests, but densely distributed urban land contributed to water quality degradation at the same buffer widths. This study could provide scientific reference for water quality management and landscape planning of lake basin in agricultural areas.