As we define the ethical framework in which directives operate, we must heed empirical research. Studies indicate that patients, wishes about life-sustaining treatments are not entirely stable over time, though choices to forgo life-sustaining treatment seem to be much more stable than choices to receive such treatment.[1] One report provides an excellent description of the vacillation and ambivalence of patients, choices when life-threatening events occur;[2] another indicates that patients vary in how strictly they wish to have their advance directives followed.[3] Very little research on the impact of advance directives upon decisions at the time of life-threatening events has been conducted. In a prospective study of advance directives that I and my collaborators conducted, several findings warranted attention.[4] First, even though the directives were created in an institutional long-term care setting and were available in the medical record, they were often left behind when patients were transferred to the acute care hospital. Second, even though the directive specified treatment choices for a variety of circumstances, including critical illness, terminal illness, and permanent unconsciousness, they were not applicable in all circumstances. The most striking finding, however, is that the advance directives were not always followed. Analysis of individual cases revealed two situations in which directives were not followed: when the preference in the directive was too restrictive to allow care that the family or provider believed to be approriate at the moment of illness, or when the family or provider did not believe the preferred care would be beneficial. Care was most likely to be consistent with previously expressed wishes if the patient remained competent, the directions were those previously expressed by the family, or the outcome event led to transfer to an acute care hospital setting where explicit discussion of treatment decisions was perhaps more likely. Anecdotal reports confirm that the existence of an advance directive may not lead to care consistent with a patient's wishes unless a surrogate vigorously asserts the validity of the directive.[5] On a positive note, one study reports that dialysis patients who had either written or verbal advance directives were much more likely than patients without directives to be reconciled to their death and to experience a death characterized by good interactions among patient, family, and medical staff., Mixed results have been reported from studies examining whether advance directives affect the cost of care at the end of life.[7] In sum, research shows that while use of advance directives may be associated with a more graceful death, these documents do not speak loudly on their own and have equivocal effects upon health care costs. While these empirical data are informative, changes in the medical approach to the end of life and die reimbursement of medical practice may have even more influence on the way we incorporate advance directives into practice. When Luis Kutner and later Sissela Bok conceived of living wills, modern medicine was tightly focused upon utilizing newly developed, life-sustaining treatments to prolong life. At the same time, the practice of medicine was reimbursed almost exclusively on a fee-for-service basis that created incentives for the physician to do more rather than less. Thus living wills were designed so that individuals who wished to forgo treatment at the end of life could fend off these tendencies. Since that time there has been a cascade of pressures upon and changes in the delivery of medical care that exert a much larger and different influence upon end-of-life decisions. New research information, demographics, economics, and ethics all have dampened the urge to use life-sustaining treatments indiscriminately at the end of life. Outcomes research has made us much more cognizant of the limited capacity of therapy to prolong life. …