This meta-analysis assessed the impact of values affirmation on the academic achievement of students under social identity threats in actual classrooms. After a systematic search yielded 58 relevant studies, multilevel analyses identified an overall affirmation effect for identity-threatened students (Hedges’ *g* = 0.15), not for identity-nonthreatened students (Hedges’ *g* = 0.01). Heterogeneity in the affirmation effect was moderate to high for identity-threatened students, with effect sizes associated with (1) a larger covariate-controlled achievement gap between nonthreatened and threatened students in the control condition, suggestive of psychological underperformance, (2) the availability of financial resources in school, (3) more distal performance outcomes, and (4) the presentation of values affirmation as a normal classroom activity rather than a research study or a non-normal classroom activity. Affirmation appears to work best when it is delivered as a normal classroom activity and where identity threat co-occurs with resources for improvement and time to await cumulative benefits.
*Keywords*: achievement gap, social identity threat, stereotype threat, values affirmation, self-affirmation, intervention