This study follows in the tradition of Cullen, Hardison, and Sackett (2004) by testing the generalizability of stereotype threat theory findings from laboratory to applied settings. Like Cullen et al., the authors developed models of the pattern of relationships between Scholastic Assessment Test (SAT) math scores and English grades that would be expected for math-identified and non-math-identified men and women if stereotype threat were operating to suppress the scores of math-identified women during SAT test administration. The study builds on Cullen et al. by creating an alternative measure of "identification" with the math domain that is premised on high school students' intention to major in math or a math-related discipline during college. Results using this alternative measure of identification were not supportive of predictions arising from stereotype threat theory, reinforcing Cullen et al.'s call for caution in generalizing stereotype threat theory lab findings to real-world testing environments.