We determined the frequency with which human immunodeficiency virus (HIV) peripheral blood mononuclear cell cultures convert from positive to negative in subjects enrolled in a substudy of AIDS Clinical Trials Group (ACTG) 320, which compared the efficacy of treatment with a combination of indinavir, zidovudine, and lamivudine (indinavir arm) to that of a combination of zidovudine and lamivudine (dual-nucleoside arm). All subjects included for study had positive baseline HIV cultures. Cultures were performed in real time with 107 fresh patient peripheral blood mononuclear cells, using the ACTG consensus method. We found lower rates of positive HIV cultures in the indinavir treatment arm than in the dual-nucleoside treatment arm (64 versus 96% at week 24, P < 0.001). Within the indinavir arm of the study, we found that positive cultures were less likely to occur in samples with a plasma HIV type 1 (HIV-1) RNA level of <500 copies/ml than in those with a level of ≥500 copies/ml (44 versus 90%, P < 0.001). In addition, HIV cultures from samples with HIV-1 RNA levels of ≥500 copies/ml turned positive 8.5 days earlier, on average, than those from samples with levels of <500 copies/ml (P < 0.001). However, 38% of samples with plasma RNA levels of <50 copies/ml still were positive for HIV by culture. Thus, the rates of HIV isolation by standard culture procedures decrease as the plasma viral load decreases below 1,000 copies/ml; however, HIV isolates were still obtained from a substantial proportion of subjects with RNA levels of <50 copies/ml. The delay in the time required for HIV cultures to turn positive should be considered when attempting to obtain an HIV isolate from patients with suppression of plasma viral load.