The US Department of Energy has conducted industrial operations at the 780 km2 Savannah River Site(SRS)in South Carolina USA since the early 1950s,which resulted in the release of heavy metals into several Savannah River tributary streams.Concern over the possible effects of these contaminants has led to the development of an ecological risk assessment program that incorporates multiple lines of evidence including comparisons of contaminant levels between reference and potentially impacted sites,the use of contaminant exposure models to estimate effects on selected ecological receptors,and bioassessments of aquatic macroinvertebrates.The contaminant exposure models estimated doses of metals ingested in fish,crayfish,sediment,and water by the river otter(Lontra Canadensis)and belted kingfisher(Ceryle alcyon)and compared these doses with chronic toxicity reference values.The bioassessments provided empirical information concerning the cumulative impacts of contaminants on lower trophic level organisms.For these analyses the SRS was partitioned spatially into Integrator Operable Units(IOUs)that corresponded to the watersheds of Savannah River tributaries.The streams within each watershed were “integrators” that received contaminants transported by surface or subsurface flow from all sources within their watersheds.Concentrations of metals were elevated in streams affected by SRS operations,but the contaminant exposure models indicated that toxicological reference values were exceeded only by aluminum and mercury.Aluminum exceedances occurred in reference IOUs without waste sites as well as potentially impacted IOUs and were likely the result of natural aluminum levels in sediments and soils.Mercury exceedances were restricted to the lower reaches of the streams where environmental conditions favored methylation and mercury bioaccumulation.Mercury levels were elevated in fish throughout the Savannah River basin as a result of the atmospheric deposition of mercury from regional sources.However,levels were higher in streams that received water from the Savannah River for reactor cooling prior to 1970 when the river was contaminated by mercury released from an industrial source upstream from the SRS.Macroinvertebrate assemblage structure was unrelated to sediment metal concentrations indicating an absence of measurable cumulative ecological effects.This study indicated that 1)modeling studies and field bioassessments provide a complementary basis for addressing the individual and cumulative effects of contaminants,2)habitat effects must be controlled when assessing contaminant impacts,and 3)sensitivity analyses of contaminant exposure models can help to apportion sampling effort and better understand processes of contaminant exposure.