Abstract Research on stereotype threat shows that students who are not socially stigmatized outperform those who are stigmatized at school. Here, we examine when and why stereotypes can become threatening in schools. We conducted a meta-analysis to measure the extent of stereotype threat effect on academic performance among elementary school children. After identifying articles written between February 2022 and 1995 (date when the first studies were carried out by Steele and Aronson), we found 17 relevant studies, with 25 independent effect sizes, including gender, ethnic origin, obesity and low socio-economic status stereotype threats. Among the moderators tested, three were significant: “Grade” (kindergarten-1st grade versus 2nd to 5th grade), “Stereotype type” (gender versus other), and “Domain” (intelligence versus math and language arts). The results showed that younger children were less sensitive to stereotype threat than 2nd to 5th graders. Contrary to our expectations, the effect of gender stereotype was not as strong as that of the other stereotypes aggregated (obesity, plus low SES and ethnic minorities). We also observed that the effect of the lower intelligence stereotype was stronger than that of lower math and language arts ability. Thereafter, we ran an analysis including all moderators, controlling the effect of each moderator by the others. The results confirm again that school age (Grade) appears to be a key determinant of the occurrence of stereotype threat. However, additional analyses cannot rule out publication bias. Consequently, it is important to conduct more studies with this population to consolidate the results obtained.