The communicative purpose of an end-of-life discussion is to change the goals of treatment for a terminal patient from therapeutic to comfort care. In this study, the authors present a comparative discourse analysis of end-of-life discussions that reached a consensus to change the goals of treatment and discussions that did not. They found that the presentation of medical information was subtly different across these discussions: Decision-making discussions were based on a consistent accumulation of negative evidence, whereas non-decision-making discussions were inconsistent in this respect, including mention of positive rather than negative outcomes of medical problems, discussion of possible treatment options, and mitigating summary statements. The authors note that end-of-life discussions with these specific features do not progress to a decision to change the goals of treatment to comfort care.