Research has shown the importance of resilience by demonstrating its significant relationship with students' academic achievement, future workplace performance, and subjective well-being. However, few studies distinguish among different definitions of resilience or distinct approaches of measuring resilience. Also there is little evidence obtained from longitudinal studies involving multiple methods in assessing resilience skills. The current study is able to overcome those limitations and make substantial progress toward the use of resilience scales and the understanding of the predictive power of resilience. Placing resilience into a broader context of non-cognitive skills, the author identifies four groups of definitions of resilience and successfully places scales of resilience into the same four categories. Using information of nearly four thousand middle school students collected longitudinally at three time points and a resilience scale which consists of three subscales, the author explores the psychometric property of the scale, asks questions on how resilience changes over time and examines the predictive validity of resilience on various future outcomes. In order to extract the true resilience variance from each of the scale and purify the scale from the wording effect, exploratory factor analysis and confirmatory bi-factor analysis are conducted. The author is able to obtain a single reliable factor which achieves scalar measurement invariance across time for each of the three subscales. However, the attempt to derive a general resilience factor fails because of the low correlations among the three subscale scores. This paper also presents the results on the change of resilience over time and the relationship between each of the resilience scores and the key outcomes. By fitting different types of hierarchical linear models and growth curve models, the author finds that resilience can significantly predict students' future grade point average and life satisfaction. The relative predictive power of different resilience scores varies by outcome. Results reveal that resilience is a promising predictor of students' academic learning and life satisfaction. Based on the results, the author provides recommendations for practitioners and researchers. Implications, limitations, and future directions of research are also discussed.