The Hydrologic Simulation ProgramFORTRAN (HSPF) model of the Ipswich River Basin previously developed by the U.S. Geological Survey was modified to evaluate the effects of the 2003 withdrawal permits and water-management alternatives on reservoir storage and yields of the Lynn, Peabody, and SalemBeverly water-supply systems. These systems obtain all or part of their water from the Ipswich River Basin. The HSPF model simulated the complex water budgets to the three supply systems, including effects of regulations that restrict withdrawals by the time of year, minimum streamflow thresholds, and the capacity of each system to pump water from the river. The 2003 permits restrict withdrawals from the Ipswich River between November 1 and May 31 to streamflows above a 1.0 cubic foot per second per square mile (ft3/s/mi2) threshold, to high flows between June 1 and October 31, and to a maximum annual volume. Yields and changes in reservoir storage over the 35-year simulation period (196195) were also evaluated for each system with a hypothetical low-capacity pump, alternative seasonal streamflow thresholds, and withdrawals that result in successive failures (depleted storage). The firm yields, the maximum yields that can be met during a severe drought, calculated for each water-supply system, under the 2003 permitted withdrawals, were 7.31 million gallons per day (Mgal/d) for the Lynn, 3.01 Mgal/d for the Peabody, and 7.98 Mgal/d for the SalemBeverly systems; these yields are 31, 49, and 21 percent less than their average 19982000 demands, respectively. The simulations with the same permit restrictions and a hypothetical low-capacity pump for each system resulted in slightly increased yields for the Lynn and SalemBeverly systems, but a slightly decreased yield for the Peabody system. Simulations to evaluate the effects of alternative streamflow thresholds on water supply indicated that firm yields were generally about twice as sensitive to decreases in the NovemberFebruary or MarchMay thresholds than to increases in these thresholds. Firm yields were also generally slightly less sensitive to changes in the NovemberFebruary than to changes in the MarchMay thresholds in the Peabody and SalemBeverly water-supply systems. Decreases in the JuneOctober streamflow threshold did not affect any of the system's firm yield. Simulations of withdrawal rates that resulted in successive near failures during the 196195 period indicated the tradeoff between increased yield and risks. The Lynn and Peabody systems were allowed to near failure up to six times. At the sixth near failure, yields of these systems increased to 10.18 and 4.43 Mgal/d, respectively; these rates increased the amount of water obtained from the Ipswich River Basin (relative to the firm-yield rate), as a percentage of average 19982000 demands, from 68 to 96 percent and from 51 to 75 percent, respectively. The SalemBeverly system was able to meet demands after the third near failure. Reservoir storage was depleted about 6 percent of the time at the withdrawal rate that caused the sixth near failure in the Lynn and Peabody system and about 3 percent of the time at the withdrawal rate that caused the third near failure in the SalemBeverly system. Supply systems are at greatest risk of failure from persistent droughts (lasting more than 1 year), but short-term droughts also present risks during the fall and winter when the supply systems are most vulnerable. Uncertainties in model performance, simplification of reservoir systems and their management, and the possibility of droughts of severity greater than simulated in this investigation underscore the fact that the firm yield calculated for each system cannot be considered a withdrawal rate that is absolutely fail-safe. Thus, the consequences of failure are an important consideration in the planning and management of these systems.