Numerous studies by Chinese and foreign scholars indicated that parents’ absence due to migration was a risk factor of such unhealthy behaviors as internet addiction, substance attachment, violence involvement, behavioral deviance and school dropout. However, relevant surveys also manifested that some left-behind children will not suffer significantly from their parents’ migration thanks to the buffering function of resilience. However, which factors determine resilience to cope and effectively reduce the risk of behavioral development is still an under-explored topic. With the help of qualitative analysis software Nvivo and statistical software SPSS, this paper performed a retrospective analysis to explore this problem based on 137 cases of adults with left-behind experiences. The cases were collected from 2017-2018 in 21 provinces. All the 137 interviewees with left-behind experiences were born between 1980 and 1999. Nvivo 11.0 was used to encode the interview materials in an open way, and SPSS 22.0 was used to transcribe the codes into quantitative data. With reference to the analysis framework of Bronfenbrenner’s Ecological Theory, this paper constructed a three-dimensional resilience model involving positive and negative factors. Non-parametric tests and partial correlation analysis were used to do basic quantitative analysis to test the relationships between behavior development and left-behind categories and home, school and social environments. The results showed that the left-behind categories were not necessarily the decisive factor in behavioral development of left-behind children. It is necessary to identify key factors from the social ecological environment where the left-behind children live. In similar social environments, family factors and school factors were all significant determinants. Relatively speaking, school risk factors and family protective factors were most critical to the behavioral development of left-behind children. The most notable school risk factor was bullying, followed by unfair treatment. Risk factors need to be avoided in both teacher-student and student-student interactions, as they could significantly contribute to negative behaviors. As for family factors, both parents and guardians could provide key protections that could influence the behavioral development of left-behind children. To be specific, family protective environmental factors could promote positive behaviors, family protective emotional factors helped to reduce negative behaviors, and both the protective and risk factors of family education had significant effects on the behavioral development of left-behind children. Therefore, families need to increase positive factors to enable children to acquire more feelings of warmth and a sense of belonging, while schools need to reduce negative factors to reduce children’s marginalization, exclusion and insecurity outside their families. From a more specific operational perspective, if we want to provide the most basic safety valve for left-behind children’s behavioral development, families should try their best to do a good job in positive tutoring, parent-child communication, social network development, and try to avoid risk factors in family instructional and emotional environments. In addition, schools should try their best to reduce risk factors among teacher-student and student-student interactions.