Resolution of long-standing debates about the role and impact of advance directives — living wills and powers of attorney for health care — has been hampered by a dearth of appropriate data, in particular data that compare the process and outcomes of end-of-life decision making on behalf of patients with and without advance directives. Drawing on a large ethnographic study of patients in two intensive care units in a large urban teaching hospital, this article compares aspects of the medical decision-making process and outcomes by advance-directive status. Controlling for demographic characteristics and severity of illness, the study finds few significant differences between patients without advance directives and those who claim to have them. Surprisingly, these few differences hold only for those whose directives are in their hospital chart. There are no significant differences between those with no directive and those claiming to have a copy at home or elsewhere. The article considers the implications if directives seemingly must be in hand to show even modest effects. Do advance directives direct? The intensive care unit data provide far more support for the growing body of literature that casts doubt on their impact than studies that promote the use of them.