Correctness of arbitral awards is a central concern in current multilateral efforts to reform investor-state dispute settlement (ISDS). Aside from protecting the disputing parties from mistakes by tribunals (retrospective correctness), corrective review also guides future interpreters not to repeat past mistakes (prospective correctness). This article assesses how effective the three existing ISDS correction mechanisms – (1) review by annulment committees or domestic courts, (2) review by the contracting parties, and (3) review by subsequent tribunals – are in promoting such prospective correctness. After assessing existing practice, the article finds that wrong decisions “don’t die”. Annulled or set-aside awards continue to be cited, contracting states’ authoritative interpretations are disregarded, and subsequent tribunals do not converge around a jurisprudence constante. This failure of corrective mechanisms to achieve prospective correctness is due to lacking legal constraints, incentives to use favorable awards even if they have been invalidated, and the simple difficulty in telling whether an award still represents “correct” law in ISDS. The article concludes by proposing possible reforms to improve prospective correctness from the shepardization of awards, to rules on precedent, and broader institutional reform.