Given the growing demand for qualified university graduates in Science, Technology, Engineering and Mathematics (STEM), significant interest exists in examining the motivating factors behind adolescents’ decisions about the educational and career pathways that they pursue in these fields. Underpinned by the expectancy-value framework, this study investigates the extent to which students’ self-concepts of ability, expectancies and values formed by Year 10 (T1) predict their STEM subject participation and preferred careers in Year 12 (T2). Contemporary, longitudinal data collected in the Study of Transitions and Education Pathways (STEPS; www.stepsstudy.org) includes participants (N= 1,172) from 9 metropolitan secondary schools in Sydney and Melbourne, Australia. A multigroup structural equation model will examine interactions between gender and STEM expectancies and values at T1, and their effects in predicting course participation and career intentions at T2. Results will provide valuable insight into the reasons students self-select out of further studies in STEM, and how to promote their career choices to continue participating in the STEM fields.