This study used data from the Seattle Social Development Project to examine factors in adolescence that affect the probability of violent behavior at age 18 among youths who received high teacher ratings of aggression at age 10. The study found a lower probability of violence among youths at age 18 was associated with attendance at religious services, good family management by parents, and bonding to school at age 15. A higher probability of later violence was associated with living in a disorganized neighborhood and having the opportunity for and involvement with antisocial peers at age 15. The likelihood of violence at age 18 among aggressive youths was reduced when they were exposed to multiple protective factors at age 15, even for those simultaneously exposed to risk factors. Implications of these findings for the development of preventive interventions during adolescence are discussed. Key words: aggression; protective factors; violence; youths ********** Studies consistently have shown that aggressive children are at high risk of later serious and chronic violent behavior (Hawkins et al., 1998). The relation between early aggression and later violence holds whether violence is measured with self-reports from youths or with data from official arrest and conviction records (Farrington, 1978, 1989). For example, Farrington (1978) found that teachers' ratings of boys' aggression at ages eight to 10 predicted later convictions for assault and robbery at ages 17 to 18. Farrington (1989) found that teacher ratings of aggression also predicted self-reports of violence in adolescence and adulthood and convictions for violent offenses up to age 32. Similarly, McCord and Ensminger (1995) found that teacher ratings of aggression for boys and girls in first grade predicted violent crime arrests to the age of 33. Nearly one-half of the aggressive boys they studied were later arrested, whereas one-third of boys without a history of aggression were arrested by that age. Although in that study fewer girls than boys were initially identified as having been aggressive in elementary school, among those who were, 17 percent later committed a violent crime compared with 7 percent of those without such ratings. More recently, Nagin and Tremblay (1999) found that trajectories of aggressive behavior beginning at age six predicted self-reports of violence and serious delinquency during adolescence among boys. Their study is important because it differentiates among forms of early conduct problems--aggression, hyperactivity, and oppositionality--as predictors of later violence. Nagin and Tremblay found that aggression predicted later violence after hyperactivity and oppositionality were controlled. Furthermore, hyperactivity and oppositionalit3, had no unique predictive effects on later violence. Their findings suggest that serious violence extends from a developmental pathway that is predominantly characterized by earlier, less serious forms of violence. Loeber and colleagues have drawn a similar conclusion (Loeber, Farrington, Stouthamer-Loeber, & Van Kammen, 1998; Loeber & Stouthamer, 1998). One explanation for the strong relation between childhood aggression and later violence points to individual traits that underlie and drive antisocial behavior. According to Moffitt's (1993) theory of life course, persistent offending, chronic patterns of antisocial behavior stem from neuropsychological impairments that affect children's acquisition of skills and development of social competencies from a young age. Without skills to regulate emotions and negotiate interpersonal conflict, individuals are prone to violence across the life span. Another explanation emphasizes processes of socialization that promote and reinforce violence as children develop. According to a social developmental perspective on the etiology of violence offered by Catalano and Hawkins (1996), behaviors formed in childhood set a foundation for future socializing experiences, which, in turn, promote (or, likewise, work against) the continuation of antisocial behavior into adolescence. …