The paper presents an impact assessment of vocational training courses, which the Piedmont Region co-financed by the ESF, both discussing the methodological feasibility and proposing a quasi-experimental evaluation strategy based on the medium-term job placement of vocational training students. The authors illustrate the operational design and the implementation of the assessment, with particular attention to the identification of a proper control sample, finally identified on the basis of the course noshows and drop-outs. The paper proposes an accurate description of gross and net impact evaluation strategies. The gross impact evaluation is carried out through an analysis of the employment outcomes in the mid-term, while the net impact is estimated through net employment differentials between the main and the counterfactual sample, as well as through a multivariate probit analysis, which investigates the effects of individual characteristics on the probability of being employed a year later the end of the course. The selection bias problem is discussed, illustrating the strategy adopted to identify and counteract its effects. In conclusion, the authors distill this research experience through a series of lessons learned about both the methods and the evaluation assessment of training effectiveness. Key words: vocational training; impact evaluation; counterfactual analysis; sampling; multivariate analysis; selection bias.