BACKGROUND: The practices of euthanasia and physician-assisted suicide remain controversial. OBJECTIVE: To achieve better understanding of attitudes and practices regarding euthanasia and physician-assisted suicide in the context of end-of-life care. DESIGN: Cohort study. SETTING: United States. PARTICIPANTS: 3299 oncologists who are members of the American Society of Clinical Oncology. MEASUREMENTS: Responses to survey questions on attitudes toward euthanasia and physician-assisted suicide for a terminally ill patient with prostate cancer who has unremitting pain, requests for and performance of euthanasia and physician-assisted suicide, and sociodemographic characteristics. RESULTS: Of U.S. oncologists surveyed, 22.5% supported the use of physician-assisted suicide for a terminally ill patient with unremitting pain and 6.5% supported euthanasia. Oncologists who were reluctant to increase the dose of intravenous morphine for terminally ill patients in excruciating pain (odds ratio [OR], 0.61 [95% CI, 0.48 to 0.77]) and had sufficient time to talk to dying patients about end-of-life care issues (OR, 0.79 [CI, 0.71 to 0.87]) were less likely to support euthanasia or physician-assisted suicide. During their career, 3.7% of surveyed oncologists had performed euthanasia and 10.8% had performed physician-assisted suicide. Oncologists who were reluctant to increase the morphine dose for patients in excruciating pain (OR, 0.58 [CI, 0.43 to 0.79]) and those who believed that they had received adequate training in end-of-life care (OR, 0.86 [CI, 0.79 to 0.95]) were less likely to have performed euthanasia or physician-assisted suicide. Oncologists who reported not being able to obtain all the care that a dying patient needed were more likely to have performed euthanasia (P = 0.001). CONCLUSIONS: Requests for euthanasia and physician-assisted suicide are likely to decrease as training in end-of-life care improves and the ability of physicians to provide this care to their patients is enhanced.