This paper investigates the joint effects of academic self-efficacy and stress on the academic performance of 107 nontraditional, largely immigrant and minority, college freshmen at a large urban commuter institution. We developed a survey instrument to measure the level of academic self-efficacy and perceived stress associated with 27 college-related tasks. Both scales have high reliability, and they are moderately negatively correlated. We estimated structural equation models to assess the relative importance of stress and self-efficacy in predicting three academic performance outcomes: first-year college GPA, the number of accumulated credits, and college retention after the first year. The results suggest that academic self-efficacy is a more robust and consistent predictor than stress of academic success.