Current recommendations about the care of dying patients advise that healthcare professionals understand and respect the goals, priorities, needs, and suffering of each dying patient and have command of the skills and resources required to address these concerns. Studies of important features of terminal illness, current use and outcome of intensive care for the terminally ill, and interventions designed to improve the outcome of care for patients who die in ICUs are reviewed to examine discrepancies between recommendations and the reality of ICU care for dying patients. Evidence indicates that it is difficult to predict the time of death or determine patient preferences about treatment prior to death. The utilization of intensive care prior to death varies widely across the United States and is a function of available resources more than individual patient need or choice. The pattern of withdrawal of life-sustaining treatment also varies widely and does not seem to follow guidelines. Families of deceased patients report that care could be improved by increased attention to analgesia and communication. The largest interventional study, the Study to Understand Prognoses and Preferences for Outcomes and Risks of Treatment (SUPPORT), which provided physicians with information about patient prognosis and preferences for care, did not alter outcomes of end-of-life care. Smaller but successful interventional studies have included examination of an alternative team that provides care tailored to the needs of dying patients, a service tailored to promote family contact with the dying patient, and proactive consultation to facilitate care planning and communication with families. Research suggests that clinicians should be cognizant of the difficulty of predicting death and anticipate the need to change the goals of care as therapeutic trials fail; anticipate and treat bothersome symptoms of dying patients; recognize that family support and contact between the dying patient and family facilitate decision-making and acceptance of death; and facilitate the coordination of care and the development of alternative care teams in order to optimize end-of-life care.