Groundwater, subsurface stormflow, and overland flow components of discharge, derived from a hydrological model that was applied to a forested headwater catchment in north central Virginia, were used with measured stream water and lysimeter concentrations of dissolved silica to investigate the hydrochemical behavior of the catchment. Concentrations in base flow, taken to be a reflection of groundwater, vary with discharge, an observation in conflict with the typical assumption of constant concentration used in end-member mixing analyses. This observed flow dependence was modeled by considering the concentration in groundwater to be related to the saturation deficit in this zone. A positive correlation between the average groundwater saturation deficit and base flow dissolved silica concentrations is consistent with batch experiments and petrographic analysis of regolith core samples, which both indicate an increase in silica available for dissolution with depth in the groundwater zone. In the absence of subsurface storm flow zone sampling during rainfall events a constant concentration was assumed for this zone. Concentration-discharge (C-Q) paths in the stream were used to evaluate the modeled stream silica concentrations. An inconsistency in the direction of the modeled C-Q rotations suggests that the storm flow zone dissolved silica concentration may also vary with time, because of the flushing of high-concentration, prevent soil water on the rising limb of the storm hydrograph. For this catchment in Virginia the assumption of a constant concentration for subsurface storm flow, as well as for base flow, appears to be invalid. ¿ 2001 American Geophysical Union