The hydrology of the basin-fill aquifer in the Tularosa Basin was evaluated through construction and calibration of steady-state and transient three-dimensional ground-water-flow simulations. Simulations were made using the U.S. Geological Survey finite-difference modular ground-water-flow computer software MODFLOW-96. The transient simulation covered 1948-2040. Both steady-state and transient simulations were calibrated by matching simulation output to available ground-water-level measurements. The root-mean-square error of the steady-state calibration in the well-calibrated area of the ground-water-flow simulation was 6.3 meters, and root-mean-square errors of individual transient-calibration points ranged from 0.8 to 17.0 meters. The areal distribution of water-level measurements used in the steady-state and transient calibrations restricts the well-calibrated area of the model to the eastern side of the Tularosa Basin. Water levels in the La Luz Creek subbasin area were underestimated by both the steady-state and transient models, suggesting that the hydrology of this area is not well represented in the model. About 143,000 cubic meters per day of recharge is estimated to enter the basin-fill aquifer from subbasins that rim the Tularosa Basin. The estimated recharge is about 4-5 percent of total precipitation in most subbasins. Approximately 88 percent of total recharge left the basin-fill aquifer as evapotranspiration under predevelopment conditions. Water levels were simulated for 1948, 1995, and 2040 under scenarios of zero and maximum return flows. Estimated return flows from municipalities were calculated on the basis of data in the Tularosa Basin Regional Water Plan for 2000-2040. Agricultural return flows were estimated primarily on the basis of ground-water-withdrawal, ground-water-depletion, surface-water-withdrawal, and surface-water-depletion data for the Tularosa Basin. The ground-water-flow simulation was sensitive to the return-flow scenario in the agricultural area near Tularosa and decreasingly sensitive to the south. Declines in simulated water levels near Tularosa between 1948 and 1995 were as large as 30 meters under the zero return-flow scenario and 15 meters under the maximum return-flow scenario. Declines in simulated water levels between 1995 and 2040 were as large as 25 meters under the zero return-flow scenario and 15 meters under the maximum return-flow scenario. Comparison of water levels measured near Tularosa in 1991 and water levels simulated under the maximum return-flow scenario for 1991 suggests that declines in simulated water levels near Tularosa may be overestimated under the zero return-flow scenario. Declines in simulated water levels near the City of Alamogordo well field between 1948 and 1995 were as large as 15 meters under the zero return-flow scenario and 10 meters under the maximum return-flow scenario. Simulated declines in water levels between 1995 and 2040 were nearly 15 meters under both return-flow scenarios assuming that all projected increases in withdrawal came from existing City of Alamogordo public-supply wells and all withdrawal from the wells came from the basin-fill aquifer. Declines in simulated water levels near the Holloman Air Force Base well fields between 1948 and 1995 and between 1995 and 2040 were less than 5 meters under both the zero and maximum return-flow scenarios. In 1995 under the zero return-flow scenario, an estimated 56,000 cubic meters of water per day was removed from aquifer storage. Of the approximately 199,000 cubic meters of water per day that left the aquifer under 1995 conditions, 40 percent left the basin-fill aquifer as ground-water withdrawal, 51 percent as evapotranspiration, 7 percent by interbasin ground-water flow into the Hueco Bolson, and 2 percent by flow into creeks and springs. Generalized directions of ground-water flow were simulated for 1948, 1995, and 2040 for much of the eastern part of the Tularosa Basin. Localized