A dual-factor model of mental health (DFM) examines the presence of wellness (i.e., subjective well-being; SWB) and psychopathology (i.e., internalizing and externalizing behavior problems) in explaining youth mental health functioning. Using a dual-factor model, previous research has yielded four unique groups: the complete mental health, the vulnerable, the symptomatic but content, and the troubled. Some researchers believe that DFM with integrated indicators may have some potential problems (Doll, 2008; Wang & Zhang, 2011), such as, lack of screening sensitivity. Affective self-regulatory efficacy is related to students’ depression and anxiety (Bandura, et al., 2003). Students’ academic self-efficacy and academic emotions contribute to their learning outcome and mental health (Dong & Yu, 2010; Dong, et al., 2014; Xiong, et al., 2011). The present empirical investigation aimed at testing the applicability of using life satisfaction on the wellness dimension and internalizing problems on the illness dimension of DFM, comparing academic and social functions of the four mental health groups, and further investigating the stability of mental health status and predictors of change in group membership. Method：Cross-sectional design and longitudinal design were applied in the present study. Consent procedure was followed in the study. This research took cluster sampling method to collect data from 1293 middle school students (Mage = 15.52, SD = ±0.75) and among them 531 were followed as the subjects of a longitudinal study. The self-report questionnaires used in the study included affective self-regulatory efficacy scale, academic self-efficacy scale, academic emotions scale, life satisfaction scale, and internalizing problems scale (Youth Self-Report). The survey was conducted at two time points: September 2013 and March 2014. Structural equation modeling was applied to test the applicability of DFM model. MANOVA and Post Hoc tests (LSD) were applied to investigate the the four groups’ academic and social functions. Repeated measures of means and multiple regressions were applied to test the predictors of change in group membership. Results：(1) The DFM based on life satisfaction and internalizing problems was feasible；(2)The complete mental health group demonstrated the best academic and social functions, and the troubled the poorest. The symptomatic but content demonstrated higher self-efficacy to regulate negative affect, academic self-efficacy and enjoyment than the troubled. The vulnerable had comparable levels of self-efficacy to regulate positive affect and enjoyment to the troubled. (3) From T1 to T2, middle school students’ mental health was relatively stable. The majority of the longitudinal sample did not change in mental health status. The proportions of positive change and negative change were close to each other. Among the change groups, the complete mental health had the highest stability, while the symptomatic but content had the lowest stability. (4) Self-efficacy for negative affect regulation, enjoyment and hopelessness predicted students’ life satisfaction and internalizing problems significantly. DFM with life satisfaction on the wellness dimension and internalizing problems on the illness dimension is applicable to middle school students in China, providing an alternative research perspective to DFM with integrated indicators. Intervention programs need to pay more attention to the symptomatic but content, since they are the least stable in mental health. Self-efficacy for negative affect regulation and academic emotions are good predictors of change in mental health group membership.