Purpose: The purpose of this paper is to analyse the effect of an Italian training program on the re-employment probability of young unemployed workers. The program consists exclusively of workplace training and is coordinated by employment centers, even if it is fully implemented by firms. Design/Methodology/Approach: The authors develop a discrete duration analysis. In particular, the authors compare the re-employment process of individuals that just finished their workplace training program with individuals that just ended their job. The authors specifically take into account the issue of self-selection adopting the propensity score matching estimation. Findings: The results suggest that this workplace training program improves only the immediate re-employability of trained workers, failing to bestow them with durable human capital improvements. These results appear to be robust to spurious duration dependence and to self-selection. The analysis focuses on unobserved heterogeneity and, accounting for it, the authors show that the training implementation is useful to divide "good" trainees (in terms of unobserved heterogeneity) from "bad" ones. Social Implications: Therefore, the authors suggest that firms are exploiting training as a screening device and that the implemented program is successful in easing the connection between workers and firms, but it fails to provide a durable improvement in skills and in re-employment prospects. Originality/Value: The evaluation of this program is important because it focus specifically on the workplace component of training, whereas previous analyses focused on generic training, because it evaluates a program targeting youth unemployment which is one the most urgent economic issues and because it helps in understanding the actual processes adopted by firms when implementing workplace training.