Some studies on untreated patients have shown a general correlation between plasma HIV copy number and plasma infectivity in in vitro models. Recent observations also indicate that HIV-RNA level is an important predictor of perinatal transmission and may also have a role in heterosexual transmission. To further analyse the correlation between HIV viral load and plasma infectivity, we studied the relationship between HIV-1 plasma copy number and plasma infectivity prior to and during treatment with antiretroviral combination regimens in HIV-1 infected adults. Plasma infectivity was assessed in vitro by coculture of plasma from HIV-positive patients with PHA-stimulated fresh PBMC from uninfected donors. A positive plasma isolation, in almost all cases (43/45) and irrespective of treatment status, was associated with an HIV viral load higher than 100000 copies per ml, with higher plasma HIV-1 RNA values in isolation-positive samples compared with isolation-negative samples (median values, 710000 vs. 37500 copies per ml, respectively). SI and NSI strains had similarly high viral load values (470000 vs. 790000 copies per ml), but CD4 counts were lower in the SI phenotype group. Our data indicate that low levels of viral load are only exceptionally associated with isolation from plasma in the in vitro model we used. This observation confirms indirectly the presence of an association between viral load and infectivity. The requisite of a high plasma viral load in order to obtain infectivity (i.e. positivity of HIV isolation from plasma) also seems maintained under antiretroviral treatment, adding confidence in the conclusion that reductions in viral load translate into reduction of plasma infectivity. Due to the extreme complexity of factors determining transmission, a very prudent interpretation of the results is essential when information from experimental studies has to be transferred to clinical situations requiring assessment of risks or clinical decisions.