Previous research showed that fear-inducing graphic warning labels can lead to cognitive dissonance and defensive responses. Less threatening, social-related warning labels do not elicit these defensive responses, making them more effective in preventing smoking in adults. Given that smoking numbers are still too high among youngsters, it is crucial to investigate how warning labels should be designed to prevent teenagers from starting smoking in the first place. In two studies, we investigated whether comparable effects of social-related warning labels could be (...) observed in a group of teenagers (14-17 years) who are not yet legally allowed to smoke. In addition, we tried to replicate earlier findings with smoking and non-smoking adults. Participants were presented with either health warning labels, social warning labels, or no warning labels. Subsequently, their explicit cognitions (i.e., risk perception, attitude towards smoking) and their implicit associations of smoking with healthiness/unhealthiness (Study 1a and Study 1b) and with positivity/negativity (Study 2a and Study 2b) were assessed. Results showed that in both studies, adult smokers had a higher risk perception and a more positive attitude towards smoking than adult non-smokers. Additionally, social warning labels lead to stronger implicit associations between smoking and negativity in Study 2 in the adult groups. In the teenage group, social warning labels lead to more positive attitudes than health warning labels in Study 2. No further effects on risk perception or implicit associations were found in the teenage group. Possible explanations are discussed. (shrink)