Unilateral medical futility policies, which allow health-care providers to limit or withdraw life-sustaining treatment over patient or surrogate objections, are increasingly designed around a procedural approach. Medical or ethics committees follow a prespecified process, the culmination of which is a justified decision about whether ongoing treatment should be withheld or withdrawn. These procedures have three stages. First, health-care providers must decide to refer patients for consideration of whether ongoing treatment is futile. Second, the committees involved must decide whether ongoing treatment is actually futile. Third, there is a clinical outcome that often is, but not always, patient death. We review the available data on procedure-based futility policies, arguing that there is limited information on their potential harms and how these harms are distributed. We consider the ethical implications of policy-making under informational uncertainty, invoking the precautionary principle—in the absence of clear data, if a policy has significant risk of significant harm, the burden of proof that it is not harmful falls on those recommending the policy—as the guiding moral standard for hospitals and professional organizations considering whether to adopt a procedural approach to medical futility. On the basis of this principle, we argue that any new futility guideline must include a significant commitment to collecting prospective data on its application.