Respect for patients' self-determination has long been considered central to efforts to improve end-of-life care, yet efforts to promote advance directives or engage patients in end-of-life discussions are often unsuccessful. In this article, I contend that this is because the shared decision-making approach typically used in healthcare assumes patients' capacity to make rational choices, which is not always possible in end-of-life decisions. Drawing on decision theory, behavioral psychology, and related studies of end-of-life care, I present a growing body of evidence that suggests the novelty, complexity, and uncertainty of end-of-life circumstances make rational and stable preferences difficult to establish. I argue that an effective decision-making approach for the terminally ill must recognize and respond to the unique characteristics of end-of-life choices, including their nonrational dimensions. I conclude with a description of an initiative that appears to do so, resulting in increased patients' satisfaction.

Copyright 2014 The Journal of Clinical Ethics. All rights reserved.