More than sixty percent of the rule-breaking acts committed by Dutch adolescents are carried out with peers. Despite abundant evidence that peer influence is a driving force in adolescent risk-taking, it remains unclear how adolescents influence each other’s behavior. What do peers do or say to encourage or discourage risk-taking? This topic fits squarely in the burgeoning integration of situational perspectives with traditional criminological theory. Also, the increasing role of online communication challenges existing knowledge of peer influence. Does peer influence operates similarly in online versus offline settings? To answer these questions, we need information about adolescents’ decision-making and adolescents’ interactions with peers as they unfold. Gaining this information requires a microsocial experimental research design and in-depth conversation analysis. In the proposed project, I will break down ‘peer influence’ into specific verbal and nonverbal cues. I will assess these cues by observing adolescents’ real-time group interactions while they play a videogame. In this game, participants make decisions under time pressure facing risk. In the offline condition, participants play the game in groups of two to three peers, who can give advice on what to do. In the online condition, peers are located in different rooms and give advice through a chat program. These online and offline interactions are recorded and coded for conversational cues, distinguishing content, direction (‘what is said’), style of the message (‘how it is said’), and the source (‘who says it’). Participants (aged 12 to 18) are recruited from middle and high schools. The findings will show what type of cues from what type of peers are most likely to lead to risk-taking, and whether these processes operate similarly in online and offline settings. These insights will constitute a significant step forward for our understanding of peer influence and will inform policies targeting adolescent rule-breaking behavior.