This article reports follow-up results during grade 9 for a multisite drug prevention program that curbed both marijuana and cigarette use during junior high. Based on the social influence model of prevention, the curriculum sought to motivate young people against drug use and to teach them skills for resisting pro-drug pressures. Thirty schools drawn from eight urban, rural, and suburban communities in California and Oregon were randomly assigned to three experimental conditions, two treatment groups and one control. Students in 20 schools received 11 lessons, 8 during grade 7 and 3 in grade 8; in 10 of the treatment schools, older teens assisted an adult teacher in program delivery. Students were pretested prior to the program (grade 7) and post-tested 24 months later (grade 9). The results show that earlier effects on cognitive risk factors (perceived consequences of drug use, normative beliefs, resistance self-efficacy, and expectations of future use) persisted through grade 9 in the teen leader schools; in the condition under which adults taught the lessons without teens, the prior beneficial effects on beliefs largely eroded. All of the earlier effects on actual use disappeared by grade 9, regardless of who taught the lessons. The authors conclude that continued reinforcement of earlier lessons may be required to sustain prevention gains through the transition to high school.