Lakes play a role as the sentinel of climate change. Surrounded by vast expanses of barren land with limited infrastructure, there is also a lack of knowledge about the dynamics of dryland lakes. The change of lake area can be effectively monitored by remote sensing, and multi-source satellite altimetry datasets provide the possibility to obtain long-term lake water level data. Using the Global Surface Water Monthly Historical dataset and altimetry water level dataset (Hydroweb), we reconstructed a time series of lake water storage changes in Xinjiang, Northwestern China, by establishing the empirical models based on the statistical relationship between the surface area and water level of each lake. We further explored lake response to climate change. The results show that the storage of water at Ayakkum Lake, Aqqikkol Lake and Aksayquin Lake have been undergoing an obvious expanding trend from 2000 to 2020, at a rate of 3.59×108m3/a, 9.43×108m3/a and 0.44×108m3/a, respectively. In the plain and transition zone, Ulungur Lake showed an upward tendency (0.413×108m3/a) in water storage, while Manas Lake and Bosten Lake experienced shrinkage with descending rates of −0.1×108m3/a and −0.86×108m3/a. Temperature changes significantly affect the lake water storage on plateaus, especially those lakes supplied with a large proportion of glacial meltwater. Precipitation is a key factor for changes of lake storage in the plain and transition zones. Meanwhile, extreme weather and man-made factors also play crucial roles. To reduce the risk of flood and drought disasters, rational regulation of water resources is required, and a large-scale integrated catchment management plan can avoid inadvertent trade-offs. This research provides a new perspective for lake water storage inversion, as well as data support for water resources management in arid areas including Xinjiang.