The current study tested the full model of the Social Cognitive Career Theory (SCCT; Lent, Brown, & Hackett, 1994, 2000) using a longitudinal sample of 1,314 Native American, Asian, and White undergraduate students majoring in science, technology, engineering, and mathematics (STEM). A series of structural equation model analyses determined that the final model offered acceptable fit to the data both in the larger sample and in sub-samples of women, men, and a combined Asian and White sample. The final measurement model was invariant across Native Americans and a combined Asian and White sample, as well. The full SCCT model did not fit well in the Native American sample, suggesting the need to identify an alternative, better-fitting structural model. Exploratory analyses identified a satisfactory model with the addition of tribal identity and removal of learning experiences. Examination of path coefficients in the full sample and Native American structural model provide support for SCCT’s main propositions, with some exceptions. Gender and racial/ethnic differences in key study variables were also identified. These findings extend research on SCCT to include a longitudinal test of the full SCCT model among an understudied student population and provide several avenues for revision/expansion of SCCT to be more compatible with students from culturally diverse background. Study limitations and directions for future research are discussed, and practical implications and suggestions for interventions are provided for increasing students’ self-efficacy, outcome expectations, interest, and intentions to pursue a STEM major based on key findings.