Depression is a prevalence psychiatric disorder in individuals exposure to trauma. Resilience and negative life events have been identified as important influence factors of depression. Studies have documented that there are complex interplays among resilience, negative life events and depression. However, few studies systematically explored the interaction relationship between the three variables. Therefore, the purpose of the present study was to explore the interplays among resilience, negative life and depressive symptoms and which circumstances those lead to steeling effect and sensitizing effect. A total of 1573 adolescents following the 2008 Wenchuan earthquake were sampled from Dujiangyan district. At the time of 6, 18 and 24 months after the earthquake, participants were asked to complete questionnaires including the Resilience Scale, Adolescent Self-Rating Life Events Checklist, Depression Self-Rating Scale for Children. 1,041 adolescents were followed up all of the three waves. Data passed the typical analysis. Reciprocal relations were found between negative life events and depressive symptoms. To be specific, Wave 1 negative life events positively predicted Wave 2 depressive symptoms (β=.12), Wave 2 depressive symptoms positively predicted Wave 3 negative life events (β=.12); Wave 1 depressive symptoms positively predicted Wave 2 negative life events (β=.20), Wave 2 negative life events positively predicted Wave 3 depressive symptoms (β=.10). Resilience and depressive symptoms also longitudinal predicted each other across waves. Specifically, Wave 1 resilience predicted Wave 2 depressive symptoms negatively (β=-.09), Wave 2 depressive symptoms negatively predicted Wave 3 resilience (β=-.08). In addition, Wave 2 negative life events negatively predicted Wave 3 resilience (β=-.09), but resilience had no direct effect on negative life events. The results of structural equation model suggested resilience, negative life events and depressive symptoms showed a dynamic interaction relationship. Resilience negatively predicted depressive symptoms (β=-.09), and then depressive symptoms could increase subsequent negative life events (β=.12); negative life events positively predicted depressive symptoms (β=.12), and then depressive symptoms could damage individuals’ resilience (β=-.08); depressive symptoms positively predicted negative life events (β=.20), and then life events could damage individuals’ resilience (β=-.09). In conclusion, resilience, life events and depressive symptoms could influence each other. The patterns of mental recovery presented two different types: steeling effect and sensitizing effect. Those had good mental health and higher level of resilience were more likely to successfully cope with life events and became better and better. In contrast, those had higher psychopathological symptoms and encountered more life events were less likely to keep mental health and high level of resilience. Adolescents had lower level of resilience, higher psychopathological symptoms and encountered higher level of life events should be provided appropriate psychology intervention.