Brazil's groundbreaking response to AIDS combines prevention strategies with the distribution of free antiretroviral therapies (ARVs). This policy was initiated in 1996 through an inventive combination of activist forces, the interests of a reforming state, international institutions, and the pharmaceutical industry. Some 170,000 people are currently taking anti-HIV drugs that are paid for by the Brazilian government. Both AIDS mortality and the use of hospital services have subsequently fallen by more than 50 percent, and this policy is widely touted as a model for stemming the AIDS crisis in the developing world.

But how do individual sufferers fare in the long term as they engage with this large-scale intervention, particularly in poor urban areas, where the epidemic is spreading most rapidly?

In 1997 and in 2001, Torben Eskerod and I worked in Caasah, a community-run AIDS hospice in the city of Salvador, in the Brazilian state of Bahia. Caasah was founded in 1992 when a group of homeless AIDS patients, former prostitutes, transvestites, and drug users squatted in an abandoned hospital formerly run by the Red Cross. Soon, if surprisingly, Caasah became a nongovernmental organization and began to receive funding from a World Bank loan disbursed through the Brazilian government. Throughout the country, other "houses of support" (casas de apoio) like Caasah mediate the relationship between AIDS patients and the haphazard, limited public health care infrastructure. However, in order to belong to these makeshift institutions of care, people must break with their old habits, communities, and routines as they forge new biographies.

"Caasah had no government," recalled Celeste Gomes, Caasah's director: "They did whatever they wanted in here. Everybody had sex with everybody, they were using drugs…. There were fights with knives, broken bottles, the volunteers had to hide in the bathroom. Police officials came here all the time, threatening to kick us out, saying that there was no command here, that all these people were marginals."

By the mid-1990s, the unruly patients in Caasah had been evicted, and a smaller version of the group began to undergo an intense process of resocialization mediated by psychologists and nurses. Patients who wanted to stay in the institution had to change their antisocial behaviors and adhere to medical treatments. Many adopted religion as an alternative value system. As Celeste put it, "With time, we domesticated them. They had no knowledge whatsoever, and we changed this doomed sense of 'I will die.' We showed them the importance of using medication. Now they have this conscience, and they fight for their lives."

With a simple chair and a black cloth against a wall, we improvised a photography studio outside Caasah's main building in March of 1997. Torben photographed each person as he or she wished to be portrayed, and I recorded their stories, past and present.

When we returned in December 2001, things had changed dramatically. Caasah had been relocated to a new, state-funded building (though it remained a nongovernmental organization). With treatment regimens available, functional residents had been asked to move out, and Caasah had been redesigned as a short-term care facility for ill patients and a shelter for HIV-positive orphans. The hospice now had a team that worked directly with local hospitals and admitted the patients that "fit into the institution and its norms," in the words of Celeste. Disturbingly, there was no systematic effort to actively track these patients and their treatment once they left.

At the state hospital I learned of the existence of a triage system (for AIDS patients), which Caasah is part of. "Homeless AIDS patients remain outside the system," one of the hospital's social workers told me. "Doctors say that they do not put these patients on ARVs, for there is no guarantee that they will continue the treatment. They are concerned about the creation of viral resistance to medication." The hospital's leading infectious disease specialist confirmed that "in theory, obviously, the doctor cannot condemn a drug user patient not to take medication . . . but the fact is that the homeless patient does not return for routine ambulatorial checkups. So what I do is to tell the patient that he has to come back. If he returns and demonstrates a strong will, we begin treatment. . . . But they never, or rarely, come back."

We looked for our former collaborators and tracked down those who had left Caasah. Some had died; others survived, married, and had children. As Torben took their portraits once again, they told us about all sorts of financial pressures, battles over discrimination, and the difficulty of obtaining access to quality health care. They told us about their will to live.