Trust and trustworthiness are important components of social capital and much attention has been devoted to their correct evaluation. In this paper, we argue that individuals' trust and trustworthiness are strongly dependent on the level of trust and trustworthiness of the social group in which subjects operate. Attitudinal indicators which are often used to measure trust and trustworthiness in economic and sociological studies are proxies of the individual's propensity to trust, but are insufficient measures of the effective level of trust since the latter may be strongly affected by the behaviour of the components of the individuals' social groups. In order to test our hypothesis, we use a rich dataset based on two experiments on the Trust Game (Berg et al.; 1995), where subjects also filled a questionnaire containing the main attitudinal questions the EVS (the European Value Survey) uses to measure individuals' trust. We then compare the ex-ante behavioural and attitudinal measures of trust with the ex post relative measures. Our main finding is that trust strongly varies once the individual is informed on the on the level of trustworthiness of the social group to which he\she has been allocated during the experiment. This difference is higher the higher is the family level of income and the parental education status of the subjects. We also find that relative behavioural measures are not correlated to attitudinal measures (Glaeser et al., 2000, Lazzarini, 2005), but they are strongly correlated to groups' trustworthiness. We also find that similar social preferences profiles (between Senders and Recipients) tend to enhance the degree of behavioural trust.