Although an increasing number of studies have demonstrated the short-term impacts of behavioral nudges to achieve public policy objectives, less is known about their longer-term impacts. In a randomized experimental design with over 100,000 households, we study the longer-term impacts of a one-time behavioral nudge that aimed to induce voluntary reductions in water use during a drought. Combining technical information, moral suasion, and social comparisons, the nudge has a surprisingly persistent effect. Although its effect size declines by almost 50% after 1 year, it remains detectable and policy-relevant six years later. In fact, the total reduction in water use achieved after the 4-month period targeted by the intervention is larger than the total reduction achieved during the target period. Further analysis suggests that the intervention works through both short-lived behavioral adjustments and longer-lived adjustments to habits or physical capital. Treatment effects are not detectable in homes from which the treated consumers have moved, which provides suggestive evidence that these longer-lived adjustments are mobile rather than incorporated into the housing stock. The persistence of the effect makes the intervention more cost-effective than previously assumed (cost drops by almost 60%). Nevertheless, water utilities may find this persistence undesirable if the nudges are intended to have only a short-run effect on demand during environmental emergencies.