The present series of studies explores the relationship between two conceptions of self-esteem: the self as experienced by the individual and the self as presented to others. Traditional measures of self-esteem are employed, and two new methods of assessing self-esteem in adolescents are introduced: an observational measure of self-esteem behavior (presented self) and a repeated self-report measure (experienced self). The presented self (observed behavior and peer ratings) and the experienced self (Rosenberg, Lerner, and a new method, paging devices) measures were consistent within themselves in assessing self-esteem. The presented self and the experienced self were not, however, significantly related to each other. We propose a moderator variable (“defensiveness”) to explain these discrepancies, and suggest that the observation of adolescent behavior provides a more accurate assessment of self-esteem than self-report measures.