Self-esteem research has been in "crisis" during the last decade, due to the lack of strong, consistent correlations between self-esteem and behavioral outcomes. Some researchers have interpreted this as indicating that self-esteem is inconsequential in many important areas of life. However, the model of direct causality used in correlational research, between a general self-esteem trait and specific behaviors, may be unrealistic. In contrast, this paper develops a model of self-esteem-motivated behaviour as originating from past, current or future (desired) self-concepts. This model shows how an interaction of catalytic factors determines how self-esteem influences behaviour. That is, what "self-esteem" actually "does." By clarifying the different ways in which self-esteem affects behavior, the model shows that construing self-esteem as a passive variable with direct causal influence on behavior is inadequate and misleading and that previous contradictory results are a consequence of this misconceptualization and subsequent reification of self-esteem. Because self-esteem and the self-concept are inseparable (one is an attitude towards the other) self-esteem-motivated behavior is always about self-construction, and thus performative. Future self-esteem research and theory should therefore focus on how people seek to enact, maintain, or defend a desired identity through performative actions.