Mentalization-based treatment in groups (MBT-G) has never been tested in adolescents with Borderline Personality Disorder (BPD) in a randomized controlled trial. The current study aimed to test the long-term effectiveness of MBT-G in an adolescent sample with BPD or BPD features (≥ 4 DSM-5 BPD criteria). Hundred and eleven patients with BPD (n = 106) or BPD features (n = 5) were randomized to either (1) a 1-year modified MBT-G program comprising three MBT introductory sessions, five individual case formulation sessions, 37 weekly MBT group sessions, and six MBT-Parent sessions, or (2) treatment as usual (TAU), defined as at least 12 individual monthly treatment sessions with follow-up assessments at 3 and 12 months post treatment. The primary outcome was the score on the Borderline Personality Features Scale for Children (BPFS-C), and secondary outcomes included clinician-rated BPD symptoms and global level of functioning as well as self-reported self-harm, depression, externalizing and internalizing symptoms, and caregiver reports. There were no statistically significant differences between MBT-G and TAU on the primary outcome measure or any of the secondary outcomes. Both groups showed improvement on the majority of clinical and social outcomes at both follow-up points, although remission rates were modest with just 35% in MBT-G and 39% in TAU 2 years after inclusion into the study. MBT-G was not superior to TAU in improving borderline features in adolescents. Although improvement was observed equally in both interventions over time, the patients continued to exhibit prominent BPD features, general psychopathology and decreased functioning in the follow-up period, which points to a need for more research and better understanding of effective components in early intervention programs. The ClinicalTrials.gov identifier is NCT02068326.