A study was conducted to review selected features of the Natural System Model, version 4.3 . The Natural System Model is a regional-scale model that uses recent climatic data and estimates of historic vegetation and topography to simulate pre-canal-drainage hydrologic response in south Florida. Equations used to represent the hydrologic system and the numerical solution of these equations in the model were documented and reviewed. Convergence testing was performed using 1965 input data, and selected other aspects of the model were evaluated.Some conclusions from the evaluation of the Natural System Model include the following observations . Simulations were generally insensitive to the temporal resolution used in the model. However, reduction of the computational cell size from 2-mile by 2-mile to 2/3-mile by 2/3-mile resulted in a decrease in spatial mean ponding depths for October of 0.35 foot for a 3-hour time step.Review of the computer code indicated that there is no limit on the amount of water that can be transferred from the river system to the overland flow system, on the amount of seepage from the river to the ground-water system, on evaporation from the river system, or on evapotranspiration from the overland-flow system . Oscillations of 0.2 foot or less in simulated river stage were identified and attributed to a volume limiting function which is applied in solution of the overland-flow equations. The computation of the resistance coefficient is not consistent with the computation of overland-flow velocity. Ground-water boundary conditions do not always ensure a no-flow condition at the boundary. These inconsistencies had varying degrees of effects on model simulations, and it is likely that simulations longer than 1 year are needed to fully identify effects. However, inconsistencies in model formulations should not be ignored, even if the effects of such errors on model results appear to be small or have not been clearly defined.The Natural System Model can be a very useful tool for estimating pre-drainage hydrologic response in south Florida. The model includes all of the important physical processes needed to simulate a water balance. With a few exceptions, these hydrologic processes are represented in a reasonable manner using empirical, semiempirical, and mechanistic relations . The data sets that have been assembled to represent physical features, and hydrologic and meteorological conditions are quite extensive in their scope.Some suggestions for model application were made. Simulation results from the Natural System Model need to be interpreted on a regional basis, rather than cell by cell. The available evidence suggests that simulated water levels should be interpreted with about a plus or minus 1 foot uncertainty. It is probably not appropriate to use the Natural System Model to estimate pre-drainage discharges (as opposed to hydroperiods and water levels) at a particular location or across a set of adjacent computational cells. All simulated results for computational cells within about 10 miles of the model boundaries have a higher degree of uncertainty than results for the interior of the model domain. It is most appropriate to interpret the Natural System Model simulation results in connection with other available information. Stronger linkages between hydrologic inputs to the Everglades and the ecological response of the system would enhance restoration efforts .