The present meta-analysis tests the bottom-up and top-down models of self-esteem, by synthesizing the available longitudinal evidence on prospective effects between global and domain-specific self-esteem. The bottom-up model assumes that people’s domain-specific self-esteem influences their global self-esteem, whereas the top-down model assumes the reverse direction of effects. Eight domains of self-esteem were assessed: academic abilities, physical appearance, athletic abilities, morality, romantic relationships, social acceptance, mathematics, and verbal abilities. We conducted a comprehensive search of the literature, which led to the inclusion of data from 43 independent samples (total N = 24,668). One-stage meta-analytic structural equation modeling was used to estimate the coefficients of interest. There was no evidence of publication bias. Overall, the results indicated a pattern of reciprocal prospective effects between global and domain-specific self-esteem. Bottom-up effects were significant in all domains except verbal abilities (mean effect sizes ranged from .05 to .19). Top-down effects were significant in all domains except mathematics (mean effect sizes ranged from .05 to .12, except .01 in the mathematics domain). None of the moderators tested (i.e., age, gender, measure, time lag, and publication year) was significant in any of the domains, which strengthens the generalizability of the results. In sum, the findings provide support for both bottom-up and top-down effects, suggesting a reciprocal relation model between global and domain-specific self-esteem. The discussion addresses the implications of the findings for research in the field of self-esteem.