Interruptions are thought to be significantly associated with medication administration errors. Researchers have tried to reduce medication errors by decreasing or eliminating interruptions. In this article, we argue that interventions are often (perhaps unreflectively) based on one particular model of risk reduction-that of barriers placed between the source of risk and the object-to-be-protected. Well-intentioned interventions can lead to unanticipated effects because the assumptions created by the risk model are not critically examined. In this article, we review the barrier model and the assumptions it makes about risk and risk reduction/prevention, as well as the model's incompatibility with work in healthcare. We consider how these problems lead to interruptions interventions with unintended negative consequences. Then, we examine possible alternatives, viz organizing work for high reliability, preventing safety drift, and engineering resilience into the work activity. These all approach risks in different ways, and as such, propose interruptions interventions that are vastly different from interventions based on the barrier model. The purpose of this article is to encourage a different approach for designing interruptions interventions. Such reflection may help healthcare communities innovate beyond old, ineffective, and often counterproductive interventions to handle interruptions.