Hospice began as a social movement outside of mainstream medicine with the goal of helping those dying alone and in unbearable pain in health care institutions. The National Hospice Study, undertaken to test whether hospice improved dying cancer patients' quality of life while saving Medicare money, found hospice care achieved comparable outcomes to traditional cancer care and was less costly as long as hospice lengths of stay were not too long. In 1982, before study results were final, Congress created a Medicare hospice benefit under a capitated per diem payment system restricting further treatment. In 1986 the benefit was extended to beneficiaries living in nursing homes. This change resulted in longer average lengths of stay, explosive growth in the number of hospices, particularly of the for-profit variety, and increases in total Medicare expenditures on hospice care. An increasingly high proportion of beneficiaries receive hospice care. However, over 30 percent are served fewer than seven days before they die, while very long stays are also increasingly common. These and other factors raise quality concerns about hospice being disconnected from the rest of the health care system. We offer suggestions regarding how hospice could be better integrated into the broader health care delivery system.