Models of receptor action are valuable for describing properties of ligand-receptor interactions and thereby contribute to mechanism-based risk assessment of receptor-mediated toxic effects. In order to build such a model for the aryl hydrocarbon receptor (AHR), binding affinities and CYP1A induction potencies were measured in PLHC-1 cells and were used to determine intrinsic efficacies for 10 halogenated aromatic hydrocarbons (HAH): 2,3,7, 8-tetrachlorodibenzo-p-dioxin (TCDD), 2,3,7, 8-tetrachlorodibenzofuran (TCDF), and eight polychlorinated biphenyls (PCB). TCDD, TCDF, and non-ortho-substituted PCBs 77, 81, 126, and 169 behaved as full agonists and displayed high-intrinsic efficacy. In contrast, the mono- and di-ortho-substituted PCBs bound to the AHR but displayed lower or no intrinsic efficacy. PCB 156 was a full agonist, but with an intrinsic efficacy 10- to 50-fold lower than non-ortho-substituted PCBs. PCB 118 was a very weak partial agonist. PCBs 105 and 128 were shown to be competitive antagonists in this system. The model was then used to predict CYP1A induction by binary mixtures. These predictions were tested with binary mixtures of PCB 126, 128, or 156 with TCDD. Both PCB 156 (a low-intrinsic efficacy agonist) and PCB 128 (a competitive antagonist) inhibited the response to TCDD, while the response to TCDD and PCB126 was additive. These data support the following conclusions: 1) only 1-2% of the receptors in the cell need be occupied to achieve 50% of maximal CYP1A induction by one of the high-intrinsic efficacy agonists, demonstrating the existence of "spare" receptors in this system; 2) the insensitivity of fish to ortho-substituted PCBs is due to both reduced affinity and reduced intrinsic efficacy compared to non-ortho-substituted PCBs; 3) PCB congeners exhibit distinct structure-affinity and structure-efficacy relationships. Separation of AHR ligand action into the properties of affinity and intrinsic efficacy allows for improved prediction of the behavior of complex mixtures of ligands, as well as mechanistic comparisons across species and toxic endpoints.