This meta-analysis synthesizes experimental studies on the immediate effects of discrimination on mental health, exploring the effects of different paradigms and discrimination types on diverse facets of mental health. We analyzed data from a systematic literature search (73 studies; 12,097 participants; 245 effect sizes) for randomized controlled trials with manipulation of discrimination as a predictor and mental health as an outcome using a three-level random-effects model. Experimentally manipulated discrimination led to poorer mental health (g = −0.30), also after controlling for publication year, region, education level, and methodological quality. Moderator analyses revealed stronger effects for pervasive (g = −0.55) compared to single-event manipulations (g = −0.25) and a trend toward weaker effects for samples with nonmarginalized (g = −0.16) compared to marginalized identities (g = −0.34). Gender and age did not moderate the effect. Discrimination had the largest effects on externalizing (g = −0.66) and distress-related outcomes (g = −0.41); heterosexism (g = −0.66), racism (g = −0.32), and sexism (g = −0.30) had the largest effects on mental health. Convenience sampling compromised generalizability to subgroups and the general population, downgrading methodological quality for all included studies. When interpreting the findings, selective samples (mostly young female adults with higher education), often limited ecological validity, and ethical restrictions of lab-induced discrimination need to be considered. These constraints likely led to conservative estimates of the mental health effects of discrimination in this meta-analysis. Future research should investigate more diverse samples, further explain the heterogeneity of findings, and explore protective factors of the effects of discrimination on mental health.