The Edwards-Trinity aquifer is a vital groundwater resource for agricultural, industrial, and public supply uses in the Pecos County region of western Texas. The U.S. Geological Survey completed a comprehensive, integrated analysis of available hydrogeologic data to develop a numerical groundwater-flow model of the Edwards-Trinity and related aquifers in the study area in parts of Brewster, Jeff Davis, Pecos, and Reeves Counties. The active model area covers about 3,400 square miles of the Pecos County region of Texas west of the Pecos River, and its boundaries were defined to include the saturated areas of the Edwards-Trinity aquifer. The model is a five-layer representation of the Pecos Valley, Edwards-Trinity, Dockum, and Rustler aquifers. The Pecos Valley aquifer is referred to as the alluvial layer, and the Edwards-Trinity aquifer is divided into layers representing the Edwards part of the Edwards-Trinity aquifer and the Trinity part of the Edwards-Trinity aquifer, respectively. The calibration period of the simulation extends from 1940 to 2010. Simulated hydraulic heads generally were in good agreement with observed values; 1,684 out of 2,860 (59 percent) of the simulated values were within 25 feet of the observed value. The average root mean square error value of hydraulic head for the Edwards-Trinity aquifer was 34.2 feet, which was approximately 4 percent of the average total observed change in groundwater-level altitude (groundwater level). Simulated spring flow representing Comanche Springs exhibits a pattern similar to observed spring flow. Independent geochemical modeling corroborates results of simulated groundwater flow that indicates groundwater in the Edwards-Trinity aquifer in the Leon-Belding and Fort Stockton areas is a mixture of recharge from the Barilla and Davis Mountains and groundwater that has upwelled from the Rustler aquifer. The model was used to simulate groundwater-level altitudes resulting from prolonged pumping to evaluate sustainability of current and projected water-use demands. Each of three scenarios utilized a continuation of the calibrated model. Scenario 1 extended recent (2008) irrigation and nonirrigation pumping values for a 30-year period from 2010 to 2040. Projected groundwater-level changes in and around the Fort Stockton area under scenario 1 change little from current conditions, indicating that the groundwater system is near equilibrium with respect to recent (2008) pumping stress. Projected groundwater-level declines in the eastern part of the model area ranging from 5.0 to 15.0 feet are likely the result of nonequilibrium conditions associated with recent increases in pumping after a prolonged water-level recovery period of little or no pumping. Projected groundwater-level declines (from 15.0 to 31.0 feet) occurred in localized areas by the end of scenario 1 in the Leon-Belding area. Scenario 2 evaluated the effects of extended recent (2008) pumping rates as assigned in scenario 1 with year-round maximum permitted pumping rates in the Belding area. Results of scenario 2 are similar in water-level decline and extent as those of scenario 1. The extent of the projected groundwater-level decline in the range from 5.0 to 15.0 feet in the Leon-Belding irrigation area expanded slightly (about a 2-percent increase) from that of scenario 1. Maximum projected groundwater-level declines in the Leon-Belding irrigation area were approximately 31.3 feet in small isolated areas. Scenario 3 evaluated the effects of periodic increases in pumping rates over the 30-year extended period. Results of scenario 3 are similar to those of scenario 2 in terms of the areas of groundwater-level decline; however, the maximum projected groundwater-level decline increased to approximately 34.5 feet in the Leon-Belding area, and the extent of the decline was larger in area (about a 17-percent increase) than that of scenario 2. Additionally, the area of projected groundwater-level declines in the eastern part of the model area increased from that of scenario 2 two individual areas of decline coalesced into one larger area. The localized nature of the projected groundwater-level declines is a reflection of the high degree of fractured control on storage and hydraulic conductivity in the Edwards-Trinity aquifer. Additionally, the finding that simulated spring flow is highly dependent on the transient nature of hydraulic heads in the underlying aquifer indicates the importance of adequately understanding and characterizing the entire groundwater system.