Objective: To examine the longitudinal association between individual subthreshold symptoms and onset of major depressive disorder (MDD) in adolescence. Method: Data for analysis come from the Oregon Adolescent Depression Project, a prospective epidemiological study of psychological disorders among adolescents, ages 14 to 18 years, from the general community. A total of 1,709 adolescents completed the initial diagnostic assessments between 1987 and 1989 (T1) and approximately 1 year later (T2), 1,507 adolescents returned for readministration of assessments (88% response). Results: After controlling for history of depression and gender, seven of the nine "DSM-III-R" symptoms of depression predicted MDD incidence when tested in separate models. Endorsement of each symptom at T1 increased the likelihood of MDD incidence between T1 and T2. A summary model that included the seven "DSM-III-R" symptoms as predictors was significant, with sad mood contributing unique variance to the prediction of MDD onset (odds ratio = 2.01). Conclusions: These findings suggest that much of the variance is shared among symptom predictors and the co-occurrence of symptoms is what constitutes the greatest risk. Moreover, the presence of sad mood contributes additional unique variance to prediction and supports the centrality of depressed mood to MDD. (Contains 3 tables.)