The objective of this study was to test the ability of existing hydrogeomorphic (HGM) functional assessment models and our own proposed models to predict rates of nitrate production and removal, functions critical to water quality protection, in forested riparian wetlands in northeastern New Jersey. In particular, the relationship between rapidly measured structural indicators and complex nitrogen (N) cycling functions was evaluated as well as the ability of generalized biogeochemical models to predict rates of specific N cycling processes. Additional models were designed specifically to predict net nitrification and denitrification rates, using both rapid and non-rapid variables based on known controlling factors on these two processes. Existing models and our own rapid models were not able to describe net nitrification and denitrification rates in urban riparian wetlands. Our additional models based on non-rapid variables, quantified through long-term hydrological monitoring and laboratory analyses of soil properties, successfully described net nitrification rates, but surprisingly failed to adequately predict denitrification rates. The high variability in hydrological and soil properties and the complexity of site-specific relationships between disturbance, altered hydrology, and hydrologically driven N cycling processes in urban wetlands restricts our ability to predict denitrification rates using simple models. It is recommended that functional assessment models be designed to describe specific, process-based functions which are based on structural indicators that are well linked to process rates. These indicators should be based on measurements from at least one year of hydrological monitoring and from simple laboratory measurements of soil properties. Detailed studies of populations of wetlands may be necessary to validate specific forms of models.