Water-column nitrogen (N) loss, driven by anammox and denitrification, accounts for ~40% of global oceanic N loss, but the relative contribution of the two pathways remains unresolved. Stoichiometry predicts that, at steady state, anammox can account for at most 29% of N loss in anoxic waters, while observations from the Eastern Tropical Pacific (ETP) show that anammox frequently dominates N loss. Knowing the relative contribution of each pathway matters because only denitrification produces the potent greenhouse gas N₂O as an intermediate. Here, we couple the regional model ROMS-BEC with a nitrogen cycle module (NitrOMZ) to resolve the contribution of both pathways to N loss across the ETP. The model reproduces the high anammox to denitrification ratio observed at the sampling locations but yields a ratio of 32:68 when integrated over the entire water column of the ETP, close to the stoichiometrically expected ratio of 29:71. Hence, the apparent mismatch between the stoichiometrically expected and observed ratios in the ETP mainly arises from spatial and, to a lesser extent, temporal sampling biases. Short-lived denitrification pulses missed by discrete sampling, previously proposed to explain the low observed contribution of denitrification, account only partially for this mismatch. Overall, the anammox to denitrification ratio is mainly controlled by oxygen and increases with oxygen concentration. In addition, we find that high anammox to denitrification ratios locally coincide with net N₂O production and low ratios with net N₂O consumption, although this relationship breaks down at larger spatial and longer temporal scales. These findings advance our understanding of N loss and its implications for N₂O production and highlight undersampled regions that may be important to monitor as oxygen minimum zones change in extent and structure under global change.