Two studies were carried out to compare verbal and behavioural measures of stealing and to investigate the effects of costs and benefits on stealing. In the first, 25 youths were interviewed and given opportunities to steal. The 10 who stole were not significantly more likely to say that they would steal in a hypothetical situation. In the second study, the behavioural experiments of Farrington and Knight (1979) on stealing from a “lost” letter were described to 10 subjects as hypothetical situations. Once again, there was little relationship between the verbal and behavioural results. Furthermore, the verbal results of the first study (the effects of costs and benefits on stealing) were not the same as those of the second. It is concluded that valid information about stealing can only be obtained in experiments in which actual stealing behaviour is studied.