Medical care at the end of life consumes 10% to 12% of the total health care budget and 27% of the Medicare budget. Many people claim that increased use of hospice and advance directives and lower use of high-technology interventions for terminally ill patients will produce significant cost savings. However, the studies on cost savings from hospice and advance directives are not definitive. The 3 randomized trials show no savings from these interventions, but either they are too small for confidence in their negative results or their intervention and cost accounting are flawed. The nonrandomized trials of hospice and advance directives show a wide range of savings, from 68% to none. Five methodological issues obscure the assessment of these studies: (1) selection bias in those patients who use hospice and advance directives, (2) the different time frames of assessing the costs, (3) the limited types of medical costs evaluated, (4) the variability of reporting the savings, and (5) the lack of generalizability of the findings to other patient populations. A more definitive study that assessed patients' end-of-life care preferences, use of hospice and advance directives, and direct and indirect costs would be desirable. In the absence of such a study, the existing data suggest that hospice and advance directives can save between 25% and 40% of health care costs during the last month of life, with savings decreasing to 10% to 17% over the last 6 months of life and decreasing further to 0% to 10% over the last 12 months of life. These savings are less than most people anticipate. Nevertheless, they do indicate that hospice and advance directives should be encouraged because they certainly do not cost more and they provide a means for patients to exercise their autonomy over end-of-life decisions.