Background: Aggression in residential youth care institutions is a frequent problem. Objective: The present short-term longitudinal study examined individual and institutional predictors of aggression in a group of 198 adolescents placed in open, semi-secure and secure residential institutions from the perspective of the importation and deprivation model. Methods: A total of 198 adolescents in residential youth care filled in questionnaires regarding group climate and aggression with a 3 month interval. Hierarchical multiple regression analyses were performed to test the degree to which individual and contextual factors predict aggression. Results: Very limited support was found for the effect of contextual factors; only repression showed a trend, predicting direct aggression, while gender composition of the living groups yielded a small effect. Girls placed in same-gender groups showed lower levels of indirect (relational) aggression compared to adolescents placed in mixed-gender or boys-only groups, even when controlled for gender and initial levels of aggression. Type of institution (i.e., level of security) did not predict differences in aggression. In particular individual characteristics of the adolescents were associated with later aggression, including initial levels of aggression, showing substantial 3 months stability, age and gender of the adolescents. Conclusions: These findings are in line with research showing that aggression is relatively stable. Very limited support for environmental effects was found.