Treatment of youth in residential care may be a challenging task for most providers because comorbid problems are common and general psychosocial functioning is low. Cognitive behavioral therapy (CBT) is found to be the most effective treatment but results in only rather small improvements. Hence, there is potential to improve treatment approaches. Acceptance and Commitment Therapy (ACT) could be one such approach. The purpose of this study was to test the effectiveness and feasibility of a brief trans diagnostic ACT group intervention for youth with comorbid problems in residential care. We also wanted to see whether increased psychological flexibility (PF) mediated potential positive outcomes, and to test the intervention under real-world conditions in residential care when delivered by less-specialized staff. With a quasi-experimental design, 69 youth (mean age 17.3 years) received Treatment-As-Usual (TAU), and 91 youth received TAU with an additional 12 h of ACT in a group setting (TAU + ACT). Follow-ups were conducted two weeks, 1 month, 6 months and 18 months after baseline. Intention-to-Treat (ITT) analyses showed statistically significant improvements 18 months after baseline, measured by the interaction of treatment and time for the primary outcomes of, anxiety [ACT * Months = -0.885 (0.445), p < 0.05, d = 0.34], depression [ACT * Months = -1.058 (0.526), p < 0.05, d = 0.39] and psychological flexibility [ACT * Months = -0.970 (0.413), p < 0.05; ACT * Months(2) = 0.053 (0.023), p < 0.05; d = 0.44] in TAU + ACT youth compared to TAU alone. Regarding secondary outcomes 6 months after baseline, the TAU + ACT group showed a significant decrease in anger, disruptive behavior, and increased self-concept, with small to medium effect sizes. We also observed that psychological flexibility mediated the decrease in the primary outcome of decreased anxiety. ACT in group format may be of help in promoting various positive outcomes for youth in residential care when added to treatment as usual. Increasing PF is a promising process variable that can be targeted to increase the effectiveness of interventions for this population.