Numerous studies have shown that riparian wetlands can play an important role in reducing nitrate concentrations before the ground water discharges into streams. Denitrification has been identified as an important process for this removal. Several approaches have been proposed to predict the denitrifying removal capacity of a riparian wetland, but no widely used tool exists to precisely quantify this capacity at the landscape scale. We propose such a methodology based on modeling the spatial variation of soil-water interactions in the entire riparian wetland. Mean values of denitrification enzyme activity (DEA) within three soil-denitrifying classes were 604, 212, and 24 ng N g(-1) h(-1) for Classes 3, 2, and 1, respectively. The study area, having a ground surface of about 15000 m2, was underlain by an aquifer with a calculated volume of 60000 m3, less than 10000 m3 of which corresponded to active denitrifying horizons (Classes 2 and 3). By volume, approximately 30% of Class 3 and 70% of Class 2 were interacting with ground water. The denitrifying removal capacity of our wetland was calculated to be about 1.8 kg N m(-2) yr(-1). The calculated denitrifying capacity of our site was less than expected. This is due to the fact that not all ground water interacts with the horizons having the highest denitrifying capacity. Thus, we show that whatever the system is, specific local pedological and hydrogeological conditions and their interactions are paramount in controlling the denitrification process.