To be an autonomous agent is to determine one’s own path in life. However, this cannot plausibly be seen as a one-off affair. An autonomous agent does not merely set herself on a particular course and then lock the steering wheel in place, so to speak, but must maintain some form of ongoing control over her direction in life—must keep her eyes on the road and her hands on the wheel. Circumstances often change in important and unexpected ways, after all, and it is reasonable to think that a crucial part of autonomy consists of the ability and disposition to recognize and properly respond to such changes. This implies, I contend, that a patient whose initial decision to undergo a given treatment satisfied plausible requirements of autonomy, but who is now unable to recognize that available evidence indicates the need to reconsider her medical situation and options has come to lack autonomy with respect to her desire to continue that treatment. However, and despite its importance with respect to both theoretical understandings of autonomy and applications of the concept to clinical ethics, this ongoing aspect of autonomy has received little attention. This paper aims to go some way toward remedying that. I first critically review two of the few theories of autonomy that do address “evidence-responsiveness” so as to identify and elaborate what I take to be the most promising way in which to account for this aspect of autonomy. After considering and responding to a possible objection to the evidence-responsiveness condition I propose, I conclude by discussing its clinical implications. That condition, I argue, is not merely theoretically sound, but can and should be applied to clinical practice.