A growing body of longitudinal studies suggests that low self-esteem is a risk factor for depression. However, it is unclear whether other characteristics of self-esteem, besides its level, explain incremental or even greater variance in subsequent depression. We examined the prospective effects of self-esteem level, instability (i.e., the degree of variability in self-esteem across short periods), and contingency (i.e., the degree to which self-esteem fluctuates in response to self-relevant events) on depressive symptoms in 1 overarching model, using data from 2 longitudinal studies. In Study 1, 372 adults were assessed at 2 waves over 6 months, including 40 daily diary assessments at Wave 1. In Study 2, 235 young adults were assessed at 2 waves over 6 weeks, including about 6 daily diary assessments at each wave. Self-esteem contingency was measured by self-report and by a statistical index based on the diary data (capturing event-related fluctuations in self-esteem). In both studies self-esteem level, but not self-esteem contingency, predicted subsequent depressive symptoms. Self-esteem instability predicted subsequent depressive symptoms in Study 2 only, with a smaller effect size than self-esteem level. Also, level, instability, and contingency of self-esteem did not interact in the prediction of depressive symptoms. Moreover, the effect of self-esteem level held when controlling for neuroticism and for all other Big Five personality traits. Thus, the findings provide converging evidence for a vulnerability effect of self-esteem level, tentative evidence for a smaller vulnerability effect of self-esteem instability, and no evidence for a vulnerability effect of self-esteem contingency.