A probabilistic risk assessment model using GOLDSIM software was developed to evaluate the uncertainty of selected hydrological and soil parameters on mercury releases from a mercury containment system, which will be constructed within the Environmental Management Waste Management Facility in the Bear Creek Valley at the Oak Ridge Reservation in Tennessee. The main objective was to determine the concentrations and risk of exceeding the drinking water standard of mercury in a selected receptor well. A series of simulations were then conducted for various design periods, with emphasis on 10,000 years to determine those concentrations and risks. Experimental data for selected parameters such as dry bulk density, partition coefficient, and porosity and infiltration rate were represented by Probability Density Functions in support of Monte Carlo analyses. A sensitivity analysis showed that concentrations and risk are, for instance, most sensitive to porosity in the unsaturated zone. The simulations suggest that all herein estimates of concentrations and risks of mercury in drinking water should be well below established limits.