supportive-expressive group, cognitive-behavioral bibliotherapy, educational brochure control). By 2-year follow-up, 14% showed onset of major/minor depressive disorders. Classification tree analysis (CTA) revealed that negative attributional style was the most important risk factor: youth with high scores showed a 4-fold increase in depression onset compared to youth who did not endorse this attributional style. For adolescents with negative attributional style, prevention condition emerged as the most important predictor: those receiving bibliotherapy showed a 5-fold reduction in depression disorder onset relative to adolescents in the three other intervention conditions. For adolescents who reported low negative attributional style scores, elevated levels of depressive symptoms at baseline emerged as the most potent predictor. Results implicate two key pathways to depression involving negative attributional style and elevated depressive symptoms in this population, and suggest that bibliotherapy may offset the risk conveyed by the most important depression risk factor in this sample.