The aim of this article is to present a theoretical model of how to undertake evaluations of prevention programs directed at adolescents. Based on the results from metaanalyses of outcome studies of school-based programs, this article concludes that interactive programs tend to be more effective than not only those mainly intended to increase factual knowledge about tobacco, alcohol, and illegal drugs, but also those initiatives aimed at influencing adolescents' attitudes toward those substances. The interactive strategy corresponds to problem-based learning or the interactive educational approach, now widely popular in educational circles. These outcome studies have provided us with important knowledge. However, there are relatively few evaluation investigations that analyze in substantive detail the process by which adolescents change their alcohol, tobacco, and illegal drug-use behavior, and alter their perceptions about these substances. In an ongoing intervention study based on an interaction framework, involving all youngsters in the ninth grade (15-16 year olds, n = 300, undertaken in year 2001-2002) in the Aland Islands, Finland, the students' perceptions regarding the risks involved in illegal drug use could not be altered. Neither was it possible to decrease the alcohol or tobacco consumption during this school year. However, what did occur was that a number of expectations regarding alcohol were changed, which, in turn, might influence the patterns of alcohol use by the target group. The ways students in the Alands celebrate the end of junior high school are culturally embedded, and this may explain the lack of success in decreasing their alcohol consumption and tobacco use at the end of the ninth grade. Based on the analysis, a combination of outcome and process evaluations is recommended in order to reach a better understanding of how school prevention programs may achieve the desired goals.