Study RegionYangtze River basin Study FocusTo improve the potential of the Gravity Recovery and Climate Experiment (GRACE) mission for estimating terrestrial water storage variations (ΔTWS) in smaller basins than its typical footprints (150,000 – 200,000 km2), an improved method named the Multi-Lagrange multiplier method (MLMM) is proposed to decrease leakage errors and restore signals from an independent hydrologic model. The MLMM is applied to estimate ΔTWS in the Yangtze River basin (YRB) and eleven of its sub-basins, and is assessed from different GRACE solutions and ground-based precipitation (P), evapotranspiration (ET), and runoff (R). Results indicate that MLMM estimates the leakage correction satisfactorily for small sub-basins. In the Tai Lake basin, the leakage ratio is improved by 16%. Excluding the Han River basin, ΔTWS from MLMM over the YRB are the best among different tested methods with a Nash-Sutcliffe Efficiency larger than 0.80, and the best root means square error (1.34 cm). New Hydrological Insights for the RegionThe main contributor to ΔTWS over the YRB is P with 60%. The contribution of ET to ΔTWS is more extensive than that of R over the middle of YRB. The ΔTWS and the total groundwater and surface water were all increasing over the YRB and eleven of its sub-basins. This paper concludes that the proposed MLMM is helpful for ΔTWS estimation in the small ungauged basins.