Terrestrial water storage consists of groundwater, soil moisture, snow, ice, surface water, and water stored in vegetation. Its importance within the Earth system is evident, but as a singular component its interactive nature and quantitative characteristics are not well understood. Current techniques for estimating changes in terrestrial water storage include ground-based observation, microwave remote sensing, water balancing, and computer modeling. The Gravity Recovery and Climate Experiment (GRACE), a satellite-based gravity mapping mission scheduled for a 2001 launch, will provide an alternative approach with distinct advantages. Redistribution of water is a primary source of variation in the gravity field over land. Therefore, it may be possible to estimate changes in terrestrial water storage based on GRACE gravity observations. The objectives of this research were to characterize terrestrial water storage variability and to explore the methods of estimation while evaluating the potential of the GRACE technique. Three studies were performed. Results demonstrate that terrestrial water storage has a strong seasonal cycle and that year-to-year variations are also significant. Multiple caveats aside, soil moisture and groundwater storage typically exhibit the greatest variability among the components of terrestrial water storage. Results also indicate that it will be possible to derive water storage variations from GRACE gravity observations on monthly and longer time steps. The primary limiting factors will be the desired spatial scale, which will be related inversely to GRACE instrument errors, and the magnitude of the variations themselves. Generally a region must be larger than 200,000 km² for the errors to be small enough to permit a meaningful estimate. Secondary limiting factors will be temporal resolution and uncertainty in the simulated fields used to remove the effect of atmospheric mass redistribution from the observed gravity signal. For a sufficiently large region, total uncertainty may be as ...