The present studies examined the association between self-concept structure and stability of self-esteem. In two daily diary studies, evaluative integration (organizing positively and negatively valenced self-beliefs into the same self-aspects) was associated with more stable self-esteem than evaluative compartmentalization (organizing positively and negatively valenced self-beliefs into separate self-aspects) among individuals with generally high self-esteem. Moreover, analyses of self-esteem reactivity confirmed that the sensitivity of state self-esteem to daily events was greater for compartmentalized individuals than for individuals with relatively integrative self-concept structures. Compartmentalization also was associated with greater sensitivity to experiences of social rejection in the laboratory, consistent with the view that integration affords greater stability of self-evaluations. These results suggest that some of the benefits believed to be associated with compartmentalization (such as high self-esteem) may have hidden costs that have not previously been considered.