Productive infection of human T lymphocytes by HIV-1 is dependent upon proliferation of the infected cell. Nonproliferating quiescent T cells can be infected by HIV-1 and harbor the virus in an inactive state until subsequent mitogenic stimulation. We use a modification of the polymerase chain reaction method, which is both sensitive and quantitative, to demonstrate that HIV-1 DNA synthesis is initiated in infected quiescent T cells at levels comparable with those of activated T cells. However, unlike that of activated T cells, the viral genome is not completely reverse transcribed in quiescent cells. Although this viral DNA structure can persist in quiescent cells as a latent form, it is labile. We discuss the lability of this HIV-1 DNA structure in relation to a "self-restricting persistent infection" by HIV-1 and propose that this may explain the low percentage of infected cells in the circulation of AIDS patients.