Families were selected from public schools when target children were in the fifth grade. Parents completed a health questionnaire consisting of 24 medical and psychiatric conditions and 34 medications. There were 185 high-risk children selected. Semistructured diagnostic interviews were conducted annually (6th through 12th grade) using the Schedule for Affective Disorders and Schizophrenia for School-Age Children-Epideminologic version (K-SADS-E)(Orvaschel et al., 1982) was used to assess children's current and lifetime history of psychopathology. Follow-up interviews were conducted through the children's senior year of high school with the Longitudinal Interval Follow-up Evaluation (Life) (Keller et al., 1987). A greater number of maternal depressive episodes predicted a more prolonged time to recovery from major depression, and greater adolescent impairment was associated with longer episode duration of both major and minor depression. The majority of significant relations between risk factors and episode durations were between characteristics of the children's disorder and time to recovery. The article suggests that treatment and prevention efforts should be directed toward offspring of parents with recurrent mood disorders, particularly adolescents with minor as well as major depression. It also recommends use of the median rather than the mean to evaluate the duration of depressive disorders. Tables, references