A theoretical framework has been developed for the evaluation of inter-ligand Overhauser effects (ILOE), predicted when pairs of ligands are observed in the presence of a macromolecular receptor which can form a ternary complex such that some of the protons on the two ligands are in close proximity with each other (generally less than approximately 5 A). Simulations for a pair of ligands with three spins each have been performed for a variety of geometric and rate parameters. Analogous to previously described calculations of TRNOE behavior, theoretical behavior of each of the nine cross peaks, A(ij), in a NOESY experiment involving ligands which can exist in the free, binary, or ternary complex states can be calculated. However, for exchange which is sufficiently rapid on the relaxation and chemical shift time scales, use of a collapsed matrix, C, corresponding to sums of sets of nine elements, will often be appropriate and generally simplifies the analysis. In order to generate inter-ligand Overhauser effects, it is optimal for the fraction of receptor involved in the ternary complex to be reasonably large; i.e., concentrations of both ligands should be near saturation. Based on a model assuming random binding order of the ligands, the dependence of ILOE resonance intensities on kinetic rate constants roughly parallels the dependence of transferred NOE (TRNOE) intensities. For diffusion controlled binding, i.e., k(on) approximately 10(8) M(-1) s(-1), the method is best suited for equilibrium dissociation constants in the micromolar-millimolar range (k(off) approximately 10(2)-10(5) s(-1)). Toward the slower dissociation rate constant end of this range, TRNOE and ILOE effects are still predicted, but the initial build-up curves become markedly nonlinear. For a kinetic binding scheme which assumes ordered binding of the ligands, the inherent asymmetry of the ligand binding process leads to more complex kinetics and alters the dependence of the ILOE on the kinetic parameters. In this case, the binding of the second ligand effectively reduces the exchange rate of the first ligand, reducing the transfer of NOE and ILOE information. The reduction in TRNOE and ILOE information which is prediced for the ordered ligand binding model is overcome at larger dissociation rate constants for either ligand 1 or ligand 2. In addition to the structural information available from ILOE data, the strong dependence of TRNOE and ILOE curves on ordered ligand binding suggests that such measurements could be useful for the characterization of ligand binding kinetics.