Water Management decisions are typically based on water demand , the available supply, and some assumptions regarding future supply. Traditionally, the future supply that is planned for is a function of the historical record, often a critical period, the drought of record, or other "worst case" scenarios. This static approach to managing hydrologic risk reflects the conservative nature of water management institutions, where the primary objective is avoiding delivery failures. In many parts of the world, however, the probability of drought conditions in any given year are influenced by climate factors that may have potential predictability. As a result, it is possible to adjust water supply decision making based on the prevailing climate conditions and associated conditional probability of drought. We apply this approach to the case of Oros Reservoir in Ceara, Brazil, where inflows are predictable with lead times of up to 18 months. Simulation results show the potential benefit of forecast-based release decisions. Surprisingly, we find that the highly conservative static risk management rules currently employed, based on an assumption of zero future inflows, reduce the reliability of water delivery compared to forecast based decision making. Results also imply that traditional water management strategies provide reliability for prioritized uses at the expense of decreased reliability for other users. Dynamic strategies that incorporate current climate conditions provide better reliability on average to all users but at the expense of increased perception of risk. We explore risk sharing strategies via economic mechanisms (e.g., insurance, compensation) to address this risk perception in an integrated robust and adaptive water management methodology.