This essay delivers two main innovations with respect to the existing literature. First, and foremost, by extending the work of Nicaise (2010) relative to the reservation wage to the case of overeducation, we propose a statistical test to discriminate between alternative theoretical interpretations of the determinants of overeducation through the Heckman sample selection procedure. Second, the essay provides the first available economic analysis of the consequences of the educational mismatch in Italy as based on AlmaLaurea data, the largest and richest data bank available in the country. The data includes a large number of university graduates enrolled in a given year before the Bologna reform and asks a large number of questions allowing us measuring among others the quality of education from high school. This wealth of information is a condition to provide the most comprehensive, accurate and reliable assessment of overeducation in the country. The educational mismatch 5 years from graduation is relatively high – at 11.4% and 8% for overeducation and overskilling, respectively – by EU standards. Ceteris paribus the parents of the mismatched have lower educational levels according to school tracking. Most humanities and social sciences degrees but also geology, biology and psychology are associated with both types of mismatch. The quality of education also correlates to the educational mismatch. We find a nonconditional wage penalty associated to overeducation and overskilling of 20% and 16% and a conditional one of about 12% and 7%, respectively. The Heckman sample selection model returns a slightly higher sample selection corrected wage penalty, supporting not only the job competition and job assignment models, but also the human capital model. Other concurrent statistical tests point to the difficulty that the educational system faces in providing work-related skills to graduates.