We evaluated the toxic-units model developed by Wildhaber and Schmitt (1996) as a predictor of indices of mean tolerance to pollution (i.e., Lenat, 1993; Hilsenhoff, 1987) and other benthic community indices from Great Lakes sediments containing complex mixtures of environmental contaminants (e.g., polychlorinated biphenyls – PCBs, polycyclic aromatic hydrocarbons – PAHs, pesticides, chlorinated dioxins, and metals). Sediment toxic units were defined as the ratio of the estimated pore-water concentration of a contaminant to its chronic toxicity as estimated by U.S. Environmental Protection Agency Ambient Water Quality Criteria (AWQC) or other applicable standard. The total hazard of a sediment to aquatic life was assessed by summing toxic units for all contaminants quantified. Among the benthic community metrics evaluated, total toxic units were most closely correlated with Lenat's (1993) and Hilsenhoff's (1987) indices of community tolerance (TL and TH, respectively); toxic units accounted for 42% TL and 53% TH of variability in community tolerance as measured by Ponar grabs. In contrast, taxonomic richness and Shannon-Wiener diversity were not correlated (P > 0.05) with toxic units. Substitution of order- or family-level identifications for lowest possible (mostly genus- or species-) level identifications in the calculation of TL and TH indices weakened the relationships with toxic units. Tolerance values based on order- and family-level identifications of benthos for artificial substrate samples were more strongly correlated with toxic units than tolerance values for benthos from Ponar grabs. The ability of the toxic-units model to predict the other two components (i.e., laboratory-measured sediment toxicity and benthic community composition) of the Sediment Quality Triad (SQT) may obviate the need for the SQT in some situations.