Soluble peptide/MHC-class-I (pMHC) multimers have recently emerged as unique reagents for the study of specific interactions between the pMHC complex and the TCR. Here, we assessed the relative binding efficiency of a panel of multimers incorporating single-alanine-substituted variants of the tumor-antigen-derived peptide MAGE-A10(254-262) to specific CTL clones displaying different functional avidity. For each individual clone, the efficiency of binding of multimers incorporating MAGE-A10 peptide variants was, in most cases, in good although not linear correlation with the avidity of recognition of the corresponding variant. In addition, we observed two types of discrepancies between efficiency of recognition and multimer binding. First, for some peptide variants, efficient multimer binding was detected in the absence of measurable effector functions. Some of these peptide variants displayed antagonist activity. Second, when comparing different clones we found clear discrepancies between the dose of peptide required to obtain half-maximal lysis in CTL assays and the binding efficiency of the corresponding multimers. These discrepancies, however, were resolved when the differential stability of the TCR/pMHC complexes was determined. For individual clones, decreased recognition correlated with increased TCR/pMHC off-rate. TCR/pMHC complexes formed by antagonist ligands displayed off-rates faster than those of TCR/pMHC complexes formed with weak agonists. In addition, when comparing different clones, the efficiency of multimer staining correlated better with relative multimer off-rates than with half-maximal lysis values. Altogether, the data presented here reconcile and extend our previous results on the impact of the kinetics of interaction of TCR with pMHC complexes on multimer binding and underline the crucial role of TCR/pMHC off-rates for the functional outcome of such interactions.