Nitrate contamination of groundwater by a sheep feedlot in Hawke's Bay, New Zealand led to closure of the feedlot in 1998. However, knowledge of the processes controlling how long the contamination will remain and an analysis of whether the current land use (vineyard) will also impact groundwater quality are required to assess long-term groundwater quality issues after feedlot operations cease. To determine the fate of NO3 following land use changes and the rate of reduction of contamination that may be expected under natural conditions following such changes, we compared the chemical concentrations of NO3–N, Cl, and alkalinity (HCO3) in the groundwater and rivers from surveys conducted in 1994 and 1995 with sampling conducted in 2001. Profile sampling of total N and C of the <2-mm size fraction in the vadose zone from two sites was used in a one-dimensional soil–plant–atmosphere system model (SPASMO) program to predict N loading in future years and predict how long it would take to improve the groundwater quality. Groundwater sampling in 1994 and 1995 determined that the highest NO3–N concentrations were under the feedlot (>140 g m−3 NO3–N) and down gradient. In 2001, 3 yr after the feedlot closed, the Cl concentrations had increased in down-gradient wells but remained similar to the 1994 survey in other wells. There has been a decrease in NO3–N concentrations in most wells, compared with the peak NO3–N concentrations recorded in the 1995 survey, but an increase compared with 1994. Alkalinity concentrations in wells located within the influence of the feedlot are approximately 150 g m−3 lower than in surrounding wells. This indicates that nitrification reactions are affecting the HCO3 concentrations in the feedlot-influenced wells. However, the HCO3 concentrations of some of these wells are increasing, indicating that nitrification could be slowing down and the aquifer is beginning to recover. SPASMO modeling indicates that NO3 contamination from the site will continue for the next 3 to 5 yr. The impact of NO3 leaching due to current land use practices is likely to be much less than the feedlot. The model predicts there will be an improvement in groundwater quality in the next 3 to 5 yr as NO3 from the feedlot eventually leaves the vadose zone profile and mixes into the unconfined aquifer.