Atmospheric deposition of biologically active nitrogen (N) has increased dramatically over the past 60 years, with far-reaching impacts on the structure and function of many ecosystems. Much research has examined the initial impacts of N enrichment; however, few studies have been multidecadal, and even fewer long-term studies have examined the longevity of N-induced impacts on N cycling after inputs cease. Here, we address this gap by reporting the state of key N pools and fluxes in a Minnesota grassland for plots that received N addition for 10 years and then none for 12 years, in comparison with plots that received annual N treatment for the entire 22 years. We found weak evidence for long-term N retention in plots that ceased receiving treatment; and in plots that continued to receive N over the 22-year period, retention that was high after 12 years (50–100% of inputs) was greatly reduced after 22 years (to 15%). In spite of this, net N mineralization rates remained elevated in plots that ceased receiving treatment 12 years prior, likely because N-rich litter maintained higher N-cycling rates. These results suggest (1) some systems do not retain much deposited N, with potentially large impacts on downstream habitats; (2) the previously reported high retention efficiencies for this and many other terrestrial ecosystems may be relatively short-lived as N sinks become saturated over time; and (3) the effects of even small amounts of retained N in N-limited environments may be particularly long-lasting. In total, these findings highlight the importance of long-term studies in evaluating the impacts of chronic N deposition to ecosystems, and urge additional research examining dynamics following N cessation to evaluate the reversibility of these impacts.