Predictions about the future course of a juvenile's behavior and how that course might be altered (positively or adversely) by particular state interventions strongly influence juveniles' dispositions. Some flaws in prediction-oriented dispositions are the vagueness of predicted behaviors, sparse evidence that recommended interventions will modify predicted behaviors, and a mixture of rationales for juvenile dispositions which frustrates definitive evaluation. Prediction is an inevitable part of juvenile dispositions, and it should be based on predictors established through research. The longitudinal studies of Wolfgang and others identify patterns of past offending as the best predictor of future offending. The use of predictors in juvenile case dispositions should be guided by an agenda. First, the intensity and duration of coercive state control over most juveniles should be reduced. Second, the continuity between juvenile and adult records should be formally established. Third, coercive resources should target juveniles whose past adjudicated behavior shows them to be violent or chronic offenders. Fourth, dispositions should be based on the desires and needs of each juvenile. 50 notes.