Abstract: Gully erosion is one of the most important drivers of land degradation, accounting for more than 50% of the total sediment production in the Loess Plateau. However, the volumetric gully erosion rate is limited during the medium to long term, due to a variety of historical information of the gullies previously unavailable. In this study, the gully growth rate over the past 53 years was estimated to compare the KH-4B imagery acquired in 1968 with the GF-7 satellite imagery in 2021. Digital Elevation Model (DEM) generated with GF-7 stereo pairs was used to calculate the gully volume. A watershed was taken as the study area, located in the first sub-region of the hilly and gully area on the Loess Plateau, well known for the severe soil erosion and high gully density. After that, 35 large gullies with the gully area ranging from 3132-34694 m2 were selected to estimate the gully linear and areal growth rates using the visual interpretation, and then to estimate gully volumetric growth rates with a volume-area relationship model. The following results were obtained. 1) There were 10 gullies undeveloped significantly, among the 35 surveyed ones. The rest of the 25 gullies increased 0.09-0.93 m/a in length, with an average value of 0.30 m/a; and the gully area increased between 0.74-51.93 m2/a, with an average value of 14.03 m2/a. 2) The growing gullies developed mainly by the head-cutting of the branches. The gully head on the shady slopes grew more quickly than those on the sunny slopes. The reason was that most gully heads on the sunny slopes were close to the ridge lines, approaching the threshold conditions of gully retreat. The terraces on the drainage area were considered to prevent the gully development, particularly for 4 of the 10 stable gullies. 3) A power function was also fitted between the volume and area of the 25 growing gullies, i.e., V=0.0866A1.4726, with R2=0.934. Each gully volume was calculated to evaluate the average gully volumetric growth rates of 161.21 m3/a, with a maximum of 723.62 m3/a. It infers that the historical KH-4B imagery was probable to extract the gully boundaries, further to estimate the gully length and area growth rates. The KH-4B imagery covering a large area in China at the end of the 1960's has been widely used to rebuild historical scenes, and then assess long term change of landscape. However, only a conservative estimate of the gully growth rate can be obtained over the past 53 years, because the 2-m resolution of the KH-4B imagery can be used to only identify the gully length growth larger than 2 times of the resolution. Meanwhile, the gully volume from the GF-7 stereoscopic satellite images can be lower than what is probably the real amount. The accuracy of the generated DEM from KH-4B stereoscopic images is not enough to extract the depth and volume of gullies. This finding can provide an important reference to estimate the gully volumetric growth rate on a large region for the medium to long periods.