It was predicted that high self-esteem Ss (HSEs) would rationalize an esteem-threatening decision less than low self-esteem Ss (LSEs), because HSEs presumably had more favorable self-concepts with which to affirm, and thus repair, their overall sense of self-integrity. This prediction was supported in 2 experiments within the ''free-choice'' dissonance paradigm-one that manipulated self-esteem through personality feedback and the other that varied it through selection of HSEs and LSEs, but only when Ss were made to focus on their self-concepts. A 3rd experiment countered an alternative explanation of the results in terms of mood effects that may have accompanied the experimental manipulations. The results were discussed in terms of the following: (a) their support for a resources theory of individual differences in resilience to self-image threats-an extension of self-affirmation theory, (b) their implications for self-esteem functioning, and (c) their implications for the continuing debate over self-enhancement versus self-consistency motivation.