Currently, the number of science, technology, engineering, and mathematics (STEM) jobs are expanding faster than the U.S. labor market; yet many individuals with STEM degrees choose to work in other fields. The present study uses social cognitive career theory as a framework for researching the impact of several variables on future expectations to continue in STEM. We measured math self-efficacy with the Patterns of Adaptive Learning Scales and impostorism from the Clance Impostor Phenomenon Scale. New measures of interest and future intentions were created. One hundred twenty-one undergraduates (47.1% women; median age = 21.00) enrolled in STEM majors completed an online survey. We hypothesized that math self-efficacy and interest would significantly predict future expectations and that impostorism would significantly add to this prediction. This theory was analyzed with a hierarchical multiple regression, and the hypothesis was supported by data from men. In the women's model, only interest emerged as a significant predictor. Implications are discussed.