The aim of this study was to explore ?theories? adolescents have about themselves, i.e. about the changes in the ways they construe themselves in the period of middle adolescence: how they construe changes and estimate them according to some dimensions relevant for change (appreciation of change, its importance, expectancy, timing, comprehensiveness, intensity and visibility in social surroundings). The study was explorative and it was performed on the sample of 96 adolescents aged 15-17 (middle adolescence). Instruments used for the study were: interview and Role Construct Repertory Grid, adjusted to the need of study. The results show that the adolescents? construct system is highly monolithic and tight in regard to constructs (black and white construing), but at the same time vague and predictively unclear in regard to elements, i. e. the stated changes, meaning that adolescents are not able to construe those changes according to certain constructs. It seems that this study included the beginning of the loosening phase (creativity cycle) in which the system is still tight and, therefore, protects from threat and anxiety, while the elements are not predictively (this is what Erikson calls moratorium, i.e. the process of active experimenting). Changes which are fully predictively are those which refer to adults' roles, such as: self confident, experienced, independent and has responsibilities, worries. On the hand, changes which are predictably the most clear to adolescents (and not appreciated) are depression and nervousness. Adolescents face an implicative dilemma: positive implications of maturity are not enough, while negative implications, such as nervousness, depression and not trusting people are clear, predictable and negative. This dilemma can be the cause of their ambivalence towards growing-up and accepting the roles of adults.