The objective of the present study was to evaluate hazard/risk assessment methods for complex environmental mixtures that involve a targeted, priority chemical approach based on the cumulative hazard/risk of known mixture components or analyses of sufficiently similar mixtures. Ten polycyclic aromatic hydrocarbon (PAH)–contaminated soils were separated into nonpolar and semipolar fractions, and both fractions elicited positive responses on the Salmonella reverse mutation assay. Targeted and nontargeted methods of hazard prediction routinely overestimated mutagenic activities for the nonpolar soil fractions, suggesting nonadditive interactions of PAHs in complex mixtures. This suggests that current risk assessment methods for complex mixtures may provide conservative estimates regarding soils contaminated with priority PAHs alone. Significant underestimations of total risk, however, will be obtained if the soils also contain unidentified PAHs as well as polycyclic aromatic compounds and related compounds that contribute to the total mutagenic activity. Furthermore, estimates of excess lifetime cancer risk associated with the nondietary ingestion of the PAH-contaminated soils studied here indicate that a traditional risk assessment model based on identified priority PAHs and an assumption of additivity generally underestimates the risk associated with the nonpolar soil fractions (in comparison to bioassay-derived risk estimates). Additional cancer risk may be associated with the more polar compounds that also are found at these contaminated sites and that rarely are included in the standard risk assessment methodology.