Abstract: Gully erosion has been considered as one of the main forms of soil erosion on the Loess Plateau of China. To control soil and water loss and improve the eco-environmental quality of the Loess Plateau, the Chinese government implemented the "Grain for Green" project in 1999 to restore vegetation on steep slope farmlands. Although the project has been implemented for two decades, soil erosion on the loess hilly region caused by extreme rainstorms is still serious. Therefore, the controlled effects of vegetation restoration on gully formation and soil erosion on loess hillslopes under extreme rainstorms need to be evaluated. Moreover, the effect of land use change on the morphology of gullies and gully volume estimation under extreme rainstorms is poorly understood. Given these issues, the Chabagou watershed (a typical watershed in the Loess Plateau of China) suffered by an rainstorm (called "7·26" rainstorm) was selected as the study area of gully erosion. Four small basins in the Chabagou watershed were selected from the upper, middle, and lower reaches, respectively, to explore the morphological characteristics and derive volume estimation models of three land uses (farmland, fallow land, and abandoned land) under extreme rainstorm. A total of 140 gullies were investigated, including 45 gullies on farmland, 66 gullies on fallow land and 29 gullies on abandoned land. The gully length was measured by the laser rangefinder and the gully was divided into three sections (upper, middle and lower) by a 50-m tape. In each section, the top width, bottom width and the depth of the gully were measured using a steel tape. The measuring times was adjusted according to the gully length to ensure the measuring accuracy. The measured cross-sectional parameters (i.e. width and depth) of the three sections were averaged as the eventual width and depth of each section, and the gully volumes of the three sections were summarized to be the amount of gully erosion. The results showed that: 1) the gullies less than 20 m on farmland, fallow land and abandoned land accounted for 55.6%, 34.8% and 44.8%, respectively. The averaged gully depth on farmland was 110 cm, which was 18.3%, and 19.2% higher than those on fallow land and abandoned land, respectively. The width-depth ratios of gully on farmland and fallow land were 0.87 and 0.84, respectively, and their cross sections were characterized by the wide and shallow shape, but the crossed section of the gully on abandoned land was closed to squared shape (width-depth ratio was 1.01). 2) The lowest gully erosion volume was found on abandoned land, which was 47.8% and 28.3% lower than that of farmland and fallow land, respectively, indicating that vegetation restoration can effectively contain gully erosion under extreme rain. 3) Gully erosion volume of farmland on upslope was largest, followed by downslope and mid-lope, while fallow land and abandoned land showed an increasing trend along the slope. For the three types of land use, the widening rate of gully on the upslope was greater than that on the downslope, but the opposite was found on the mid-slope. 4) The significant (P<0.001) power functions could express the relationships between gully erosion volume and gully length and/or gully cross-sectional area on farmland, fallow land, and abandoned land, and the cross-sectional area was a more effective parameter for estimating gully volume. This study results can provide important information for predicting the gully erosion of different land use types and controlling gully erosion in the hilly area of the Loess Plateau.