Adolescents (n=218) from 15 residential centers received an intensive nine-session HIV prevention program. The residential centers were randomly assigned to three groups: skills training, discussion-only, and control group. The two experimental groups participated in nine sessions delivered over a 3-week period. The substantive content of the skills training and discussion groups were identical, and both were based on cognitive-behavioral principles; however, the methods of delivering the material were different. The outcome variables included measurement of adolescents' knowledge, attitudes, reported intentions to cope with AIDS-risk situations, and behaviors related to AIDS. To assess adolescents' report of engagement in unsafe activities, five types of risky sexual behaviors were examined. For all of the behaviors, the respondents stated how often they had engaged in each activity in the last 3 months. Results show that one intervention model, discussion groups, produced a long-term increase in knowledge about AIDS and higher reported intentions to cope with AIDS-risk situation; however, neither skills- training nor discussion groups produced a long-term reduction in the level of engagement in high-risk behaviors. Several reasons for these results are discussed, with emphasis on an explanation based on the theoretical perspective of life chances or life options. 2 tables and 21 references