Despite great institutional efforts to recruit students, including those from underrepresented groups, into the science, technology, engineering, and mathematics (STEM) pipeline, the number of students choosing STEM-related careers remains low. Using data from the Educational Longitudinal Study, we estimated a two-level model of high school sophomores' choice of STEM-related careers with gender, math self-efficacy, socioeconomic status (SES), school type, and urbanicity as predictors. Our findings show that notwithstanding the obvious private school advantage for attaining a STEM career, attainment is still lower for female students with high math self-efficacy. The gender gap persists among private and public schools regardless of students' math self-efficacy level. Students are significantly more likely to choose a STEM-related career if they are male in a private school with high SES and math self-efficacy. School urbanicity is also a significant predictor for students having a STEM-related career. We discuss this study's implications and future research directions.