Based on satellite observations of Earth's time variable gravity field from the Gravity Recovery and Climate Experiment (GRACE), it is possible to derive variations in terrestrial water storage, which includes groundwater, soil moisture, and snow. Given auxiliary information on the latter two, one can estimate groundwater storage variations. GRACE may be the only hope for groundwater depletion assessments in data-poor regions of the world. In this study, soil moisture and snow were simulated by the Global Land Data Assimilation System (GLDAS) and used to isolate groundwater storage anomalies from GRACE water storage data for the Mississippi River basin and its four major sub-basins. Results were evaluated using water level records from 58 wells set in the unconfined aquifers of the basin. Uncertainty in the technique was also assessed. The GRACE-GLDAS estimates compared favorably with the well based time series for the Mississippi River basin and the two sub-basins that are larger than 900,000 km2. The technique performed poorly for the two sub-basins that have areas of approximately 500,000 km2. Continuing enhancement of the GRACE processing methods is likely to improve the skill of the technique in the future, while also increasing the temporal resolution. © Springer-Verlag 2006.