Imprisonment is often said to be the ultimate back-up sanction for offenders who do not abide by their non-custodial sentence. From a standard consequentialist perspective this is morally justified, if it is a cost-effective means to crime prevention. In contrast, the use of imprisonment as a back-up is much harder to justify from retributivist perspectives, with their emphasis on just desert or deserved censure. The crux is this: if the reason for a non-custodial sentence is that a prison sentence risks being a disproportionate or inappropriate sanction, retributivists need to explain how a prison term can be warranted as the backup sanction for those who breach the requirements of their non-custodial sentence, even though their original crime wasn’t serious enough to warrant imprisonment in the first place. The aim in this paper is to critically assess the extent to which retributivists can justify the use of imprisonment as the ultimate back-up sanction. In doing so, I first examine two broad strategies that are discussed in the literature, and which retributivists could employ in order to respond to this problem. The first strategy stresses how desert has only a limited role in sentencing such that it demarcates a range of deserved punishment. On this view, associated with limiting retributivism, one could initially opt for a less harsh yet deserved punishment, leaving room for the imposition of back-up sanctions when needed. The second strategy focuses on how the act of breach is a reprehensible act that can allow for a penalty increase, and thereby lead to imprisonment. Although it is argued that both strategies fail, the paper proposes an alternative solution to this problem.