Background: Community studies with 1-3 years of follow-up have reported four to five times higher mortality in HIV-2-infected than in uninfected adults. In a cohort study of HIV-1, an increasing difference in mortality rates of HIV-1-infected and uninfected individuals is expected over time, because of rising mortality with advancing HIV-1 infection. We therefore investigated long-term survival of HIV-2-infected adults. Methods: Adults enrolled in 1987 in a community study of HIV-2 infection in Guinea-Bissau were followed up with serological surveys in 1989 and 1992. Survival was assessed in 1995, 9 years after enrolment. Findings: The annual incidence of HIV-2 was 0.7% for adults and tended to be higher for older individuals than for participants aged 15-44 years (relative risk 3.21 [95% CI 0.91-11.37]). With control for age, HIV-2-infected adults had twice as high mortality as uninfected individuals (mortality ratio 2.32 [1.18-4.57]); the mortality ratio was highest in the first year of the study (4.50 [1.31-15.43]). The difference between infected and uninfected individuals was stronger for adults under 45 years of age (mortality ratio 4.72 [1.86-11.97]) than for older people (1.35 [0.51-3.56]). HIV-2-infected individuals living with an infected spouse had significantly higher mortality than HIV-2-infected individuals living with an uninfected spouse (p = 0.027). Interpretation: HIV-2-associated mortality is not increasing with length of follow-up. Mortality in HIV-2-infected adults is only twice as high as that in uninfected individuals. In the majority of adults, HIV-2 has no effect on survival.