This paper aims at estimating early labour market returns (i.e. participation probability, employment probability and log hourly earnings) of Italian university graduates across college subjects. We devote great attention to endogenous selection issues using alternative methods to control for potential self-selection associated with the choice of the degree subject in order to unravel the causal link between college major and subsequent outcomes in the labour market. We use both a propensity score matching-average treatment on the treated method and the polychotomous selectivity model introduced by Lee (1983) to investigate the existence of unobserved heterogeneity. Our results suggest that "quantitative" fields (i.e. Sciences, Engineering and Economics) increase not only participation to the labour market and employment probability but also early earnings, conditional on employment.