The mechanisms by which HIV induces the immune dysfunction clinically defined as AIDS have been a subject of intense study since the discovery of the virus in the early 1980s. Initial virological analysis demonstrated low levels of virus replication in infected patients, suggesting that the virus alone was incapable of inducing AIDS and that additional factors must also play a role in determining the HIV-associated immunologic dysfunction. This concept has recently been emphasized from a statistical perspective by the observation that while the level of HIV replication is significantly correlated with the risk of disease progression [1], this parameter only predicts a minor part of the variation in the rate of progression among infected patients [2]. In the mid 1990s, improvements in the techniques available to detect HIV demonstrated that virus replication was active throughout the course of the disease [3, 4]. In addition, the observation that inhibition of viral replication with antiretroviral drugs substantially attenuates disease progression established very clearly that virus replication is responsible for pathogenicity. What remains poorly defined, however, is the mechanistic linkage between virus replication and the onset of AIDS.

A model of CD4 T-cell depletion based entirely on direct virus infection and killing of these cells was put forward in the mid 1990s [5, 6]. This so-called ‘tap-and-drain’model proposed that progression to AIDS in HIV-infected individuals resulted from a failure of the immune system's homeostatic response to keep up with a high rate of loss of CD4 T cells [5, 6]. The model offered an explanation for the rapid increase of CD4 T-cell counts following inhibition of virus replication by antiretroviral therapy. However, this model and its later versions [7, 8] were challenged on theoretical and experimental grounds as they did not appear to grasp the complexity of T-cell dynamics in response to ongoing viral replication and painted a simplistic picture of AIDS pathogenesis [9–17].