The study was based on Mischel's social cognitive learning theory. Major personal resources for resilience were expected in the following areas: behavior construction competencies (intelligence and temperament), coding strategies and personal constructs (coping styles and constructs of social resources), behavior- and stimulus- related expectations and evaluations (self-efficacy and learned helplessness), and self-regulating systems (self- esteem and achievement motivation). Adolescents who had high, cumulative risk loads in childhood were differentiated according to groups that either showed emotional and behavioral disorders in the sense of deficit models or managed to develop in relative mental health. German adolescents between 14 and 17 years of age were randomly selected from 60 residential homes for children and adolescents. Both adolescents and home staff participated in interviews and written surveys to assess biographical risk loads and risk conditions. An analysis of group differences showed that resilient adolescents proved to be more intelligent than deviants facing the same risk. The variable of verbal intelligence, more closely related to education, played a less important role than the reasoning component. Resilients tended to cope with problems more actively than passively and fatalistically. In their self-related cognitions, they experienced themselves as being less helpless, and they had a more positive self-evaluation. In the domain of temperament, resilients proved to be more strongly approach-oriented and flexible. In addition, resilients consistently reported a larger network of social support and were more satisfied with the support they received. Resilients reported a childrearing climate that was more strongly oriented toward autonomy, more open, and less conflictive. Limitations of the study findings are noted. 56 references, 3 tables, and 2 figures