Two studies examined whether test item content might play a role in stereotype threat. In Experiment 1, male and female students in grades 7-8 completed math story problems that were identical but used either male names, female names, or featured names not gender specific (control). Performance was significantly better when males names were used, but students with lower math scores were more likely to correctly answer items with female names. There were no significant effects due to students' gender. Experiment 2 used a similar method but focused on male and female college students and introduced a stereotype threat manipulation. Specifically, half the students were told that "This test has been found to show gender differences in math performance. Overall, males have obtained higher scores than females on this test" (stereotype threat for females), or "We are simply interested in how well Canadian university students can solve these word problems" (control). Results showed that men performed better than women, but only in the stereotype threat condition. Moreover, the effect did not occur for high-ability women, and there were minimal effects associated with gender-labeling of items. These data show that children, but not young adults, were affected by the gender-labeling of math problems. However, young adults (specifically young women), were affected by a manipulation of stereotype threat with lower math performance in conditions in which gender stereotypes were invoked.