In our article titled “Compromised Autonomy and the Seriously Ill Patient,”1 in a recent issue of CHEST (April 2010), we tried to provide a careful and subtle solution to an important and vexing problem: How can clinicians properly respect the autonomy of their seriously ill patients when the capacity for autonomous choice can be fluctuating in and out of existence and can be compromised by decreased cognitive functioning, delirium, systemic weakness, pain, depression, and anxiety? The principle of patient self-determination or patient autonomy is too blunt an instrument for the complex predicaments in which such patients find themselves. When there is reason to believe that autonomy is compromised in these ways and when the patient’s wishes go against the grain of his or her best interests, physicians can only respect autonomy if they take the patient’s current wishes as but one reason in the mix of reasons that bear on making the right decision. We went out of our way to make it clear that these decisions could not be made easily, that the patient’s long-term goals and values will form part of that mix of reasons, and that explicitness and transparency were essential so that the reasoning could be open to challenge from the patient and loved ones. That is, we want nothing to do with the argument Rady and Verheijde attribute to us: that clinicians can simply override a patient’s goals and values, without trying to understand them, if an agreement about treatment cannot be reached. But simply accepting and acting on a statement of preference from a patient with compromised autonomy does not do that patient a service. Rather, an independent assessment of the best interests of the patient gives clinicians a starting point from which to evaluate claims and requests from patients and surrogates alike.