As part of a study of the feasibility of recovering freshwater injected and stored underground in south Florida, a digital solute-transport model was used to investigate the relation of recovery efficiency to the variety of hydrogeologic conditions that could prevail in brackish artesian aquifers and to a variety of management alternatives. The analyses employed a modeling approach in which the control for sensitivity testing was a hypothetical aquifer considered representative of permeable zones in south Florida that might be used for storage of freshwater. Parameter variations in the tests represented possible variations in aquifer conditions in the area. The applicability of the analyses to south Florida limestone aquifers required the assumption that flow nonuniformities in those aquifers are small on the scale of volumes of water likely to be injected, and that their effect could be represented as hydrodynamic dispersion. Generally, it was shown that a loss of recovery efficiency is caused by (1) processes causing mixing of injected freshwater with native saline water (hydrodynamic dispersion), (2) processes causing the more or less irreversible displacement of the injected freshwater with respect to the well (buoyancy stratification, background hydraulic gradients, and interlayer dispersion), or (3) processes causing injection and withdrawal flow patterns to be dissimilar (directionally biased well-bore plugging, and dissimilar injection and withdrawal schedules in multiple-well systems). Other results indicated that recovery efficiency improves considerably with successive cycles, providing that each recovery phase ends when the chloride concentration of withdrawn water exceeds established criteria for potability (usually 250 milligrams per liter), and that freshwater injected into highly permeable or highly saline aquifers (such as the 'boulder zone') would buoy rapidly. Many hydrologic conditions were posed for model analysis. To have obtained comparable results with operational testing would have been more costly by orders of magnitude. The tradeoff is that the validity of results obtained from computer modeling is somewhat less certain. In particular, results must be qualified with observations that (1) the complex set of processes lumped as hydrodynamic dispersion is represented with a somewhat simplified mathematical approximation, and (2) other flow processes in limestone injection zones are as yet incompletely understood. Despite such reservations, the study is considered a practical example of the use of transport models in ground-water investigations.