A school-based sex education program was developed in order to prevent sexually transmitted diseases and unwanted pregnancies. A Solomon four-group design, with random assignment to the different conditions, was used to evaluate an intervention based on cognitive social learning theory and social influence theory. The main goal of the intervention was to increase use of condoms. A stratified sample of 124 classes (2,411 students) was drawn at random from all the upper secondary schools (high schools/colleges) in one county in Norway. The results indicate a consistent interaction between pretest and intervention, which seems to have an effect on condom use. Pretest or intervention alone did not contribute to this effect. The interaction effect appeared among the students with few sexual partners. Several possible explanations to the observed interaction effect and the implication for future interventions are discussed.