The objective of this study was to quantify subsurface nitrogen fluxes between a riparian forest and a 4th-order mountain stream, McRae Creek, for each season of the year and during storms. A network of wells was installed on a gravel bar and a portion of the adjacent floodplain between 1989 and 1992. Water samples were collected to monitor dissolved nitrogen concentrations Advected channel water and ground water were enriched in nitrogen relative to the stream; thus, subsurface flow was a net source of nitrogen to the stream in all seasons of the year and during both base-flow periods and storms. Estimates of the flux of advected channel water and the discharge of ground water were combined with changes in mean nitrogen concentrations along subsurface flow paths to estimate nitrogen inputs to the stream. Discharge of ground water from the conifer-dominated floodplain was the largest source of nitrogen added to the stream; however, more than 50% of this nitrogen was dissolved organic nitrogen. In contrast, two-thirds of the nitrogen from the alder-dominated gravel bar was inorganic. Net nitrogen fluxes from the gravel bar to the stream were lowest during the summer when water table elevations were low. Net fluxes of nitrogen from the gravel bar to the stream were largest during the fall, especially at peak flow during storms when interstitial water in the gravel bar was enriched in NO3-. The estimated annual flux of nitrogen from the riparian forest to McRae Creek was 1.9 g/m(2) of streambed, of which 1.0 g/m(2) was inorganic. Estimated net annual flux was large relative to the estimated input of nitrogen in litterfall, or the nitrogen required to support estimated rates of primary productivity.