Studies on diverse populations have produced three clusters of factors which are predictive of future criminality. These parent factors include the parents' criminality, lack of parental supervision, conflict between the parents, and the use of harsh and physical techniques of child rearing. The factors relating to the children themselves include male gender, black race, lower I.Q., aggressive or impulsive temperament, and younger age of onset of delinquency. The school factors include interpersonal difficulties and academic difficulties. However, only half of those identified as being at high risk of future criminality will later prove to be criminal. Thus, both the costs and the benefits of early interventions must be considered. The two main costs of early intervention programs are considered to be labeling effects and the widening of the net of social control, although their magnitudes are unknown. In addition, the positive effects of intervention programs are unknown, but in the past have tended to be minimal. Thus, public policy regarding early intervention should be approached with caution. The ability to predict seems adequate for voluntary programs, but not for coercive ones. Twenty-seven reference notes are provided.