BACKGROUND: Predicting the expected outcome of a combination exposure is critical to risk -assessment. The toxic equivalency factor (TEF) approach used for analyzing joint effects of dioxin-like chemicals is a special case of the method of concentration addition. However, the TEF method assumes that individual agents are full aryl hydrocarbon receptor (AhR) agonists with parallel dose-response curves, whereas many mixtures include partial agonists. OBJECTIVES: We assessed the ability of generalized concentration addition (GCA) to predict effects of combinations of full AhR agonists with partial agonists or competitive antagonists. METHODS: We measured activation of AhR-dependent gene expression in H1G1.1c3 cells after application of binary combinations of AhR ligands. A full agonist (2,3,7,8-tetrachlorodibenzo-p--dioxin or 2,3,7,8-tetrachlorodibenzofuran) was combined with either a full agonist (3,3 ,4,4 ,5-penta-chloro-biphenyl), a partial agonist (2,3,3 ,4,4 -pentachlorobiphenyl or galangin), or an antagonist (3,3 -diindolylmethane). Combination effects were modeled by the TEF and GCA approaches, and goodness of fit of the modeled response surface to the experimental data was assessed using a nonparametric statistical test. RESULTS: The GCA and TEF models fit the experimental data equally well for a mixture of two full agonists. In all other cases, GCA fit the experimental data significantly better than the TEF model. CONCLUSIONS: The TEF model overpredicts effects of AhR ligands at the highest concentration combinations. At lower concentrations, the difference between GCA and TEF approaches depends on the efficacy of the partial agonist. GCA represents a more accurate definition of additivity for mixtures that include partial agonist or competitive antagonist ligands.