Water scarcity is a growing concern spurred by the climate crisis, and food production is a significant contributor to water use. One effective way to address water scarcity is thus to make consumers aware of how food products differ in the amount of water used in their production. How well do people know how much water is used to produce food products? And how can people’s estimation accuracy and food choices be improved? In an online study (N = 102), participants estimated the water used to produce various food products in liters. Then they received either the actual water use (seeding), a simple rule to estimate water use (rule), or they read an unrelated text (control). Then they provided estimates for previously judged as well as new food products. Finally, participants indicated which of four shopping lists of food products they considered to be associated with the lowest water use. Participants’ initial estimates were rather poor, underestimating water use by 2,652 liters on average. The rank correlation of the estimates with the actual water use was merely r = .17. Seeding improved estimates and ranking accuracy relative to the control condition, while the rule improved only the ranking accuracy. Shopping list selection was most accurate in the seeding condition and worst in the control condition. Our results demonstrate that a simple intervention improves people’s estimation accuracy and enables them to make more sustainable food choices.