||A 16,000 acre-foot reservoir is proposed to be located about 25 miles east of Grand Junction, Colorado, on a tributary of the Colorado River that drains the Sulphur Gulch watershed between De Beque and Cameo, Colorado. The Sulphur Gulch Reservoir, which would be filled by pumping water from the Colorado River, is intended to provide the Colorado River with at least 5,412.5 acre-feet of water during low-flow conditions to meet the East Slope's portion of the 10,825 acre-feet of water required under the December 20, 1999, Final Programmatic Biological Opinion for the Upper Colorado River. The reservoir also may provide additional water in the low-flow period and as much as 10,000 acre-feet of water to supplement peak flows when flows in the Colorado River are between 12,900 and 26,600 cubic feet per second. For this study, an annual stochastic mixing model with a daily time step and 1,500 Monte Carlo trials were used to evaluate the probable effect that reservoir operations may have on water quality in the Colorado River at the Government Highline Canal and the Grand Valley Irrigation Canal. Simulations of the divertible flow (ambient background streamflow), after taking into account demands of downstream water rights, indicate that divertible flow will range from 621,860 acre-feet of water in the driest year to 4,822,732 acrefeet of water in the wettest year. Because of pumping limitations, pumpable flow (amount of streamflow available after considering divertible flow and subsequent pumping constraints) will be less than divertible flow. Assuming a pumping capacity of 150 cubic feet per second and year round pumping, except during reservoir release periods, the simulations indicate that there is sufficient streamflow to fill a 16,000 acre-feet reservoir 100 percent of the time. Simulated pumpable flows in the driest year are 91,669 acre-feet and 109,500 acre-feet in the wettest year. Simulations of carryover storage together with year-round pumping indicate that there is generally sufficient pumpable flow available to refill the reservoir to capacity each year following peak-flow releases of as much as 10,000 acrefeet and low-flow releases of 5,412.5 acre-feet of water. It is assumed that at least 5,412.5 acre-feet of stored water will be released during low-flow conditions irrespective of the hydrologic condition. Simulations indicate that peak-flow release conditions (flows between 12,900 and 26,600 cubic feet per second) to allow release of 10,000 acre-feet of stored water in the spring will occur only about 50 percent of the time. Under typical (5 of 10 years) to moderately dry (3 of 10 years) hydrologic conditions, the duration of the peak-flow conditions will not allow the full 10,000 acre-feet to be released from storage to supplement peak flows. During moderate to extremely dry (2 of 10 years) hydrologic conditions, the peak-flow release conditions will not occur, and there will be no opportunity to release water from storage to supplement peak flows. In general, the simulated daily background dissolved-solids concentrations (salinity) increase due to the reservoir releases as hydrologic conditions go from wet to dry at the Government Highline Canal. For example, the simulated median concentrations during the low-flow period range from 417 milligrams per liter (wet year) to 723 milligrams per liter (dry year), whereas the simulated median concentrations observed during the peak-flow period range from 114 milligrams per liter (wet year) to 698 milligrams per liter (dry year). Background concentration values at the Grand Valley Irrigation Canal are generally only a few percent less than those at the Government Highline Canal except during dry years. Low-flow reservoir releases of 5,412.5 acre-feet and 10,825 acre-feet were simulated for a 30-day period in September, and low-flow releases of 5,412.5 acre-feet were simulated for a 78-day period in the months of August through October. In general, these low-flow releases resulted in changes to salinity concentrations ranging from slight decreases to slight increases in dissolved-solids concentrations over the range of hydrologic conditions simulated. Low-flow releases of 5,412.5 acre-feet of water over the 78-day period resulted in percentage increases in salinity greater than the measurement error for salinity in fewer than 10 percent of the driest years simulated. Low-flow releases of 5,412.5 acre-feet of water over the 30-day period coupled with peak-flow releases of as much as 10,000 acre-feet of water also resulted in percentage increases in salinity greater than the measurement error for dissolved-solids in fewer than 10 percent of the driest years simulated. Observed trends in stream dissolved-solids concentrations at the Grand Valley Irrigation Canal are similar to observations of simulated dissolved-solids concentrations change at the Government Highline Canal, however, the magnitude of percent and absolute change is less except under very dry hydrologic conditions. In addition to dissolved-solids concentration, understanding instream changes in selenium concentration following reservoir releases are of concern because selenium can be toxic to fish and other biota. In general, instream selenium concentrations are an order of magnitude greater in tributary creeks like Sulphur Gulch (1 to 25 micrograms per liter) than in the Colorado River (0.3 to 0.7 microgram per liter). Stochastic modeling indicates that random sampling may result in a 1-percent and 35-percent chance, respectively, of exceeding Colorado instream acute (18.4 micrograms per liter) and chronic (4.6 micrograms per liter) water-quality standards in Sulphur Gulch runoff. The lack of selenium in water pumped from the Colorado River to storage likely will result in diluting reservoir concentrations to respective levels ranged from 0.37 to 1.48 micrograms per liter under wet and dry hydrologic conditions. Therefore, based on the simulations and inherent assumptions, selenium concentrations in the proposed reservoir are expected to be less than the acute and chronic standards.