After several weeks of combination antiretroviral therapy, plasma HIV-1 RNA falls below the detectable range of existing assays. This decline has prompted optimists to think that HIV-1 can eventually be eradicated. Downward trends in plasma viral load during the first months of therapy were extrapolated to give estimates that 2· 3 to 3· 1 years of suppressive therapy might lead to eradication of HIV-1 if there were no tissues providing sanctuary for the virus. 1 The search for tissue sanctuaries focused on lymphoid tissues, such as tonsils, gutassociated lymphoid tissue, and lymph nodes, where most of the total body load of HIV-1 resides, 2 and on tissues that are isolated from drug-laden blood, such as brain and testes. Fortunately, HIV-1 RNA load in lymphoid tissue3 and semen4 decays in parallel with viraemia, and preliminary analysis of other tissues is also encouraging. These studies indicate that plasma viral load is a good marker of HIV-1 response to antiviral therapy in the whole body, and that the lymphoid tissues need not be biopsied for routine clinical assessment. Failure to identify tissue reservoirs of HIV-1 also bolstered hopes of eradication. The extensive search for tissue sanctuaries has, however, revealed a reservoir of HIV-1—latently infected CD4 T lymphocytes in the blood. 5–7 HIV-1 has long been known to establish latent infection in cells cultured in the laboratory. Like all complex retroviruses, HIV-1 preserves ample genetic resources, encoded at the ends of the viral genome (ie, the long-terminal repeats), which are involved in latency; this point suggests that latency confers a selective advantage on HIV-1. Indeed, latent HIV-1 infection was thought to account for the persistence of HIV-1 during the clinically silent period. 8 A shift in thinking occurred with the demonstration that rapid viral turnover underlies more than 99% of HIV-1 viraemia, 9, 10 which suggests that HIV-1 persistence might be due to continuous active replication rather than viral latency. These studies did not assess the number of latently infected cells, which may be large even if the amount of virus produced is small. Furthermore, even a small reservoir of latent infection would be sufficient to trigger viral replication if antiretroviral therapy were stopped.

HIV-1 proviral DNA can persist in blood lymphocytes despite antiretroviral therapy sufficient to suppress plasma viral RNA. Because it has a half-life of 150 days, 9 cell-associated proviral DNA can be found for years after plasma viral RNA has been suppressed below detectable limits. Fortunately, most proviral DNA in blood cells is