The current study will examine the effectiveness of a short video (i.e. vlog) intervention to promote physical activity among adolescents. In addition, this study will test whether a social network vlog interventions is more effective than a mass media vlog intervention. A specific subset of adolescents (i.e. influence agents) will be asked to create six vlogs in which they promote physical activity, based on the behavior change mechanisms proposed by the Fogg’s behavior model. Thirty classes (N=1250) will be allocated via cluster randomization to one of the three conditions. Participants will receive a research smartphones to watch the vlogs, fill out the sociometric nominations and measure the covariates. An accelerometer (Fitbit Flex) is used to measure daily physical activity. At baseline, peers will nominate each other on sociometric questions (e.g. friendship) and the most central adolescents will be approached, on the last day of the baseline measures, to become an influence agents. During the intervention week, participants receive a vlog per day of the influence agents in their class (social network condition), from influence agents of another school (mass media condition) or no vlogs (control condition). Follow-up measurement will be five weeks after the intervention. Mixed models will be used to test the effectiveness of the interventions, controlling for clustering of the data within classes, participants and days. This study will be the first social network intervention that uses vlogs to promote physical activity in adolescents and compare this to a mass media campaign and control condition.