Abstract: In order to investigate the characteristics, spatial and temporal distribution patterns and evolution patterns of agricultural water stress in China, this paper constructs the agricultural water stress index model to measure the agricultural water stress across the country, and analyzes the spatial and temporal distribution patterns with the help of visualization tools, and analyzes the spatial differences, dynamic evolution and spatial correlation patterns by using the Dagum Gini coefficient decomposition, Kernel density estimation and Moran index model. The results show that:（1）the national agricultural water stress index decreased from 2.13 to 1.59 from 2005 to 2021, and the regional stress index ranked from high to low as east ＞ west ＞ central ＞ northeast；（2）the stress index of nine regions, including Beijing, Tianjin, Hebei, Shanxi, Henan, Shandong, Jiangsu, Xinjiang and Ningxia, is greater than 1 during the period under examination, and the agricultural water stress index is roughly the spatial distribution pattern of decreases “from north to south” and “from northeast to southwest”；（3）Dagum decomposition results show that the overall Gini coefficient of China′s agricultural water stress level only decreases slightly, and the contribution of hypervariable density is the main source of the total difference；（4）The Kernel density estimates show a dynamic evolution pattern of “decreasing crest, moving to the left, shortening the right tail, and increasing the width”；（5）The spatial Moran index has been significantly positive since 2014, and is mainly distributed in the first, second, and third quadrants. The study concludes that although China’s agricultural water stress index is declining, it is still high overall; the unsustainable agricultural water resources are mainly concentrated in North China, Bohai Sea Economic Zone and Northwest China, and the inter-regional（intra-regional） stress index shows obvious unbalanced evolution characteristics; the spatial clustering characteristics of agricultural water stress mainly show high clustering（16.67% of areas）, low clustering（50% of areas） and low spatial clustering（50% of areas）. The spatial agglomeration of agricultural water resources pressure mainly shows three patterns of high agglomeration（16.67% area）, low agglomeration（50% area） and low agglomeration（30% area）.