This paper shows how discontinuities in the process of drug delivery enable but also underdetermine the isolation of a culprit in adverse medication events. A case example illustrates how we are forced to abandon conceptualizations of blame that assume a dichotomy (either culpable or not), and shift instead to a more nuanced version that estimates the degree to which an actor desired, generated, or could have foreseen the harmful outcome, and the extent to which constraints external to the actor altered the event. The paper concludes that meaningful safety interventions in a system as diverse, socially embedded and complex as health care delivery cannot just build on "good science" (e.g., good methods) to generate "root" causes. Rather, they need to somehow be sensitive to how and which narratives of success and failure are created, as these constrain which countermeasures are likely to be encouraged, funded, and accepted.