DescriptionThe research described in this report was inspired by a perspective that runs counter to a long tradition of viewing the ideal self as a collection of desirable traits. In the first study, a large cross-sectional sample of adults was asked to describe their ideal selves ("at my best") and their undesired selves ("at my worst") using an open-ended format. As anticipated, most participants described their ideal selves in terms of past, present, or future role successes, not traits. Also as predicted, undesired selves were predominantly described in terms of role failures. Sex differences with respect to undesired selfdescriptions are reported with males predominantly focusing on failed agentic roles and females focusing on both agentic and communal role disruptions. Given the role specificity used to describe ideal and undesired selves, a second study was designed to challenge the assessment method that accompanies Self-Discrepancy Theory. Some research participants completed the standard Selves-Questionnaire (SQ) by rating abstract targets (Ideal-, Ought-, and Undesired-Selves) while other participants rated selfgenerated Ideal, Ought, and Undesired roles. The SQ method predicted more robust affect scores than did the role specificity measure with respect to ideal-self discrepancies, but the reverse was true for the role specific undesired self measure. These results are followed by a discussion of their implications.