Shared decision making is inadequate in intensive care units. Decision aids can improve decision making quality, though their role in an intensive care units setting is unclear. We aimed to develop and pilot test a decision aid for shared decision makers of patients undergoing prolonged mechanical ventilation. Intensive care units at three medical centers. Fifty-three surrogate decision makers and 58 physicians. We developed the decision aid using defined methodological guidelines. After an iterative revision process, formative cognitive testing was performed among surrogate-physician dyads. Next, we compared the decision aid to usual care control in a prospective, before/after design study. Primary outcomes were physician-surrogate discordance for expected patient survival, comprehension of relevant medical information, and the quality of communication. Compared to control, the intervention group had lower surrogate-physician discordance (7 [10] vs. 43 [21]), greater comprehension (11.4 [0.7] vs. 6.1 [3.7]), and improved quality of communication (8.7 [1.3] vs. 8.4 [1.3]) (all p<.05) post-intervention. Hospital costs were lower in the intervention group ($110,609 vs. $178,618; p=.044); mortality did not differ by group (38% vs. 50%, p=.95). Ninety-four percent of the surrogates and 100% of the physicians reported that the decision aid was useful in decision making. We developed a prolonged mechanical ventilation decision aid that is feasible, acceptable, and associated with both improved decision-making quality and less resource utilization. Further evaluation using a randomized controlled trial design is required to evaluate the decision aid's effect on long-term patient and surrogate outcomes.