A watershed is a topographically defined area such that all the precipitation falling into the area leaves in a single stream. Most watersheds used in ecosystem studies are relatively small first order streams with drainage areas of 10-50 ha. Studies on the watershed level integrate all processes occurring within the boundaries of a given watershed. This may include several plant species, several soil types, contrasting microclimates, and complex hydrologic flow paths. Watershed studies provide ecosystem level data which incorporate all ecosystem processes, but does not distinguish between individual processes or their relative importance. Because of its integrating nature, watershed data can be extrapolated to a larger scale with appropriate precautions.

Watershed studies are based on input-output budgets for nutrients and water . Water and nutrient inputs are measured by collecting and quantifying water and nutrient concentrations in precipitation. Watershed outputs are quantified by measuring streamwater flow volumes using a stream gauging station. An important assumption for watershed studies is that there is negligible seepage of water to deep groundwater drainage so that all water leaves the watershed as streamflow. The difference between precipitation inputs and streamwater can then be assigned to evapotranspiration (ET = Precipitation - Streamflow). Nutrient concentrations in streamwater are measured on samples collected throughout the year to capture variations associated with differences in stream discharge and seasonality. By coupling streamwater nutrient concentrations (mg/L) with stream flow (L/s), it is then possible to calculate nutrient losses for the watershed (kg/ha/yr). The difference between nutrient inputs and outputs is a measure of the net nutrient balance associated with all pools and fluxes. Thus, it is not possible to quantify individual processes from watershed studies.

Watershed manipulations are a powerful approach for studying the effects of forest management practices on nutrient cycling processes. Manipulation studies often use the paired watershed approach in which two watersheds with similar characteristics are employed. Prior to inducing a treatment, the hydrologic and nutrient response of the paired watersheds should be monitored to verify that the two watersheds function similarly and can be considered pairs. One watershed of the pair is maintained in its original state to serve as a reference while the other watershed undergoes a particular treatment. The paired watershed approach has been used to examine the effects of timber harvest practices (e.g., whole-tree harvesting, bole-only clearcut, block cutting), fertilization, and acidification to simulate atmospheric deposition.