The freshwater content at the Mackenzie shelf/estuary is analyzed using observed and numerically simulated data. Field measurements cover the period from September 1986 to September 1987; calculations are based upon a time-dependent, one-dimensional model that treats the shelf as a single reservoir coupled to the surroundings through parameterizations of the inflows and outflows. The system is assumed to be controlled by transient Ekman flow dynamics, and forced by winds when the sea is ice free, and by tides when ice covered. Buoyancy fluxes due to river inflow, freezing and melting of ice, and outflow are added to the salinity conservation equation from which the freshwater content is calculated. Sensitivity studies from simulation of the 1986--1987 period show that the freshwater content of the Mackenzie shelf/estuary is highly influenced by freezing, ice advection off the shelf, and the wind-driven transport, all of which work effectively to remove fresh water from the shelf. Ice keels in the inner shelf may also reduce the export of fresh water to the outer shelf. The sensitivity of the system to altered river inflow from either hydroelectric development or climate change is also examined. For example, upstream storage would increase winter inflows and thus decrease the shelf's capacity to ventilate the halocline, while inflow reduction would enhance shelf ventilation. ¿ American Geophysical Union 1994