Since the late 1980s, mandatory welfare-to-work programs in the United States (US) issued sanctions on welfare beneficiaries who failed to comply with program guidelines. While numerous studies have focused on the impact of sanctions policies on desired outcomes, including higher participation rates and improved labour market outcomes, only a few studies have examined the role that commitment devices play in the success of sanction policies. This paper provides a theoretical explanation for the importance of commitment devices. It does so by applying Dixit and Nalebuff’s theory of strategic moves to the relationship between government and welfare recipients. The paper analyses how the relationship between the government and the welfare recipients changes in situations with credible vs. non-credible threat of sanctions. By doing so, it argues that commitment devices are a critical yet a largely unstudied factor in the impact of sanction policies. Further, the paper applies this theory to the US welfare-to-work programs and identifies the key commitment devices - the administrative method of auto-posting and the use of sanctioning rates as a reputation mechanism - used by the government to change its relationship with the welfare recipients. Finally, it offers a novel research method to assess whether commitment devices used in US programs are as critical to the success of sanctions policies as the theory suggests.