Consumptive use of freshwater is of concern in many estuarine ecosystems, and various frameworks have been used to prescribe environmental flows to benefit native species. However, few of these frameworks explicitly examine the potential trade-offs between socioeconomic and conservation-oriented values. This is exemplified in California, USA, where freshwater management has been an area of focus and controversy. Operations of numerous reservoirs and water diversion facilities distributed throughout the state, while critical for economic and public health benefits, have contributed to the decline of many native species. The endangered delta smelt (Hypomesus transpacificus) is endemic to the Sacramento-San Joaquin Delta, the heart of California's complex water conveyance system. To aid recovery of delta smelt, fall-timed freshwater pulse flows were implemented, which require water to be either released from reservoirs, or made unavailable to export for consumptive uses. Previous research has indicated that the effectiveness of the current pulse flow action could be improved by reconsidering the timing and magnitude; however, uncertainties in the predicted fish response to flow pulses may hinder decision-making about flow management. Using a water resource planning model, different iterations of an individual-based life cycle model, and decision analysis tools, we assessed the importance of sources of uncertainty to hypothetical flow management decisions, including uncertainty surrounding the predicted responses in delta smelt population growth rates, and variability of decision-maker's values. We found both the choice of which (if any) flow action to take for delta smelt, and the expected value of further research, depended on how decision-makers weight the delta smelt and water supply objectives. There was expected value of information (VOI) only if a decision-maker weighted the delta smelt objective ≥0.59, and within this range, research to improve estimates of changes in delta smelt prey items related to flow actions could be prioritized over other sources of uncertainty to improve outcomes of decision-making. Our study demonstrates how uncertainty, even if large, may not be equally relevant to different decision-makers (e.g., with different agency missions), and how VOI analysis can be used to guide management in an overallocated water system such as California.