This study directly compared mortality risk in homeless and nonhomeless mentally ill veterans and compared mortality rates in these groups with the general U.S. population. The study used a retrospective cohort design to assess mortality over a 9-year period in homeless (N = 6,714) and nonhomeless (N = 1,715) male veterans who were treated by Department of Veterans Affairs specialized mental health programs. The study showed that mortality rates in all homeless members of the cohort were significantly higher than the general U.S. population. Relative to nonhomeless cohort members, significant increases in mortality risk were observed in cohort members who at baseline were age 45 to 54 and had been homeless 1 year or less (RR = 1.55, 95% CI = 1.02, 2.36) and those age 55 and older who had been homeless 1 year or less (RR = 1.83, 95% CI = 1.33, 2.52). Similar, but nonsignificant trends were observed in cohort members who had been homeless more than 1 year at baseline. Additionally, medical problems at baseline and history of prior hospitalization for alcohol problems elevated mortality risk. Employment at baseline and minority group membership reduced mortality risk. The study suggests that mentally ill veterans served by specialized VA mental health programs are at elevated risk of mortality, relative to the general population. Homelessness increases this risk, particularly in older veterans, and this difference does not abate after entry into a health care system.