How to efficiently allocate scarce water resources to sustain agriculture and protect endangered species in droughts is a pressing but challenging problem. Water buybacks, a form of payment for ecosystem services where government and non-government agencies pay farmers to reduce irrigation, have been used to reallocate water from agricultural to environmental use. Due to a lack of rigorous consideration and evaluation of cost-effectiveness, little is known about whether water buybacks effectively allocate limited funds to achieve environmental improvements. This study experimentally examines the cost-effectiveness of reverse auction mechanisms to buy back irrigation water for endangered aquatic species conservation during droughts. We designed and conducted pre-registered induced-value laboratory experiments with university students and field experiments with farmers where the available funding was nearly half a million dollars. Results suggest that an auction target based on environmental benefits is more cost-effective than an acre-based target. Different irrigation reduction approaches involve trade-offs between the program’s cost-effectiveness, farmers’ earnings, and flexibility regarding drought conditions. This research advances knowledge and policy implications on designing cost-effective payments for ecosystem services with the goal of providing sufficient streamflow while also minimizing the economic impact on the agricultural sector.