We analyze the short- and long-term effects of the U.S. Vietnam-era military service on veterans' health outcomes using a restricted version of the National Health Interview Survey 1974-2013 and employing the draft lotteries as an instrumental variable (IV). We start by assessing whether the draft lotteries, which have been used as an IV in prior literature, satisfy the exclusion restriction by placing bounds on its net or direct effect on the health outcomes of individuals who are non-veterans regardless of their draft eligibility (the "never takers"). Since we do not find evidence against the validity of the IV, we assume its validity in conducting inference on the health effects of military service for individuals who comply with the draft-lotteries assignment (the "compliers"), as well as for those who volunteer for enlistment (the "always takers"). The causal analysis for volunteers, who represent over 75% of veterans, is novel in this literature that typically focuses on the compliers. Since the effect for volunteers is not point-identified, we employ bounds that rely on a mild mean weak monotonicity assumption. We examine a large array of health outcomes and behaviors, including mortality, up to 40 years after the end of the Vietnam War. We do not find consistent evidence of detrimental health effects on compliers, in line with prior literature. For volunteers, however, we document that their estimated bounds show statistically significant detrimental health effects that appear 20 years after the end of the conflict. As a group, veterans experience similar statistically significant detrimental health effects from military service. These findings have implications for policies regarding compensation and health care of veterans after service.