The main purpose of the present study was to determine the role of family flexibility, parents’ level of self-differentiation, and parental attachment style in predicting adolescents’ self-differentiation and attachment style. The statistical population of the study consisted of all Tehrani male and female high school students of the first, second and third grades. Among them, 283 students were selected via random multi-stage cluster sampling method and participated in the study alongside their parents. The instrumentation comprised Family Flexibility Questionnaire (Shakeri, 2003), Differentiation of Self Inventory (short form; Drake, 2011), and Adult Attachment Scale (Griffin & Bartholomew, 1994). Correlation matrix and stepwise regression were utilized to analyze the obtained data. Results showed that family flexibility, mother's dismissing attachment style, and mother’s emotional reactivity could positively predict 34% of variance in adolescents’ self-differentiation, whereas fathers’ preoccupied attachment style could negatively predict this much variance. In addition, it was uncovered that mother’s emotional cutoff, family flexibility, and father’s emotional cutoff could positively predict 15% of variance in adolescents’ secure attachment style, whereas father’s preoccupied attachment style could negatively predict this much variance. Family flexibility had the most important role in predicting adolescents’ insecure attachment style. Hence, it can be concluded that family flexibility, personal characteristics of parents, and the quality of interactions in family are among the most important determinants of adolescents’ self-differentiation and secure attachment style.