Notwithstanding an impressive research tradition on key skills, no clear statistical criterion exists that is suitable to determine which skills may be considered key skills. This contribution proposes one possible methodology that can be used to identify key skills. Proposing an economic definition of the key skill concept and disentangling the direct and indirect effects of skills on productivity, we develop an empirical criterion for the identification of key skills. We apply this methodology to a dataset of employed vocational education graduates. We find that problem-solving skills, independence, oral presentation/speaking skills, accuracy/carefulness and initiative/creativity may be considered key skills.