We compare the output of an 18-box geochemical model of the ocean with measurements to investigate the controls on both the mean values and variation of nitrate δ15N and δ18O in the ocean interior. The δ18O of nitrate is our focus because it has been explored less in previous work. Denitrification raises the δ15N and δ18O of mean ocean nitrate by equal amounts above their input values for N2 fixation (for δ15N) and nitrification (for δ18O), generating parallel gradients in the δ15N and δ18O of deep ocean nitrate. Partial nitrate assimilation in the photic zone also causes equivalent increases in the δ15N and δ18O of the residual nitrate that can be transported into the interior. However, the regeneration and nitrification of sinking N can be said to decouple the N and O isotopes of deep ocean nitrate, especially when the sinking N is produced in a low latitude region, where nitrate consumption is effectively complete. The δ15N of the regenerated nitrate is equivalent to that originally consumed, whereas the regeneration replaces nitrate previously elevated in δ18O due to denitrification or nitrate assimilation with nitrate having the δ18O of nitrification. This lowers the δ18O of mean ocean nitrate and weakens nitrate δ18O gradients in the interior relative to those in δ15N. This decoupling is characterized and quantified in the box model, and agreement with data shows its clear importance in the real ocean. At the same time, the model appears to generate overly strong gradients in both δ18O and δ15N within the ocean interior and a mean ocean nitrate δ18O that is higher than measured. This may be due to, in the model, too strong an impact of partial nitrate assimilation in the Southern Ocean on the δ15N and δ18O of preformed nitrate and/or too little cycling of intermediate-depth nitrate through the low latitude photic zone.