Predictive, bioassay, and field methods were compared with regard to assessing conditions and risks to terrestrial biota at a Superfund site contaminated with pesticides. Research indicated that predictive methods could result in large uncertainties, especially when attempting to estimate food chain transfer from soil to higher trophic levels (birds and mammals). Of the examined predictive methods, the soil-to-earthworm model yielded results that agreed best with field observations and literature values. Laboratory and in-field bioassays on soils provided good information on the potential toxicity of soils and the potential for bioaccumulation. In-field bioassay tests with earthworms were useful for defining the spatial distribution of toxic soils at the site as well as for indicating potential bioaccumulation in invertebrates. An attractive feature of in-field bioassays is that samples do not have to be removed from the field and assessments can begin to be made in the field. The importance of performing reality checks by conducting field studies was clearly demonstrated in our study with regard to the soil invertebrate community as well as birds and mammals. The main advantage of the field studies is that they provide direct information on conditions and thus aid in the interpretation or application of laboratory data. Difficulties were encountered in the conduct of population studies of the birds and mammals. Therefore, although site surveys of birds and mammals provide gross indications as to the presence and diversity of animals, it is probably more useful to rely on measures of residues and biomarkers to assess potential exposures and the possibility of subtle effects. Overall, it appears that ecological risk assessments at Superfund sites are best conducted by using a mix of predictive, laboratory, and field methods.