The predictive value of two methods for measuring HIV RNA concentration in plasma was assessed in relation to CD4 lymphocyte counts during the asymptomatic period of infection. The design was a retrospective longitudinal case-control study for a mean period of 60 months involving 20 asymptomatic patients included in the French National prospective survey. The CD4 counts in these patients during the last 36 months of the study were stable (non-progressors) or declined (progressors). Plasma RNA concentrations were determined in each subject annually using the AMPLICOR and NASBA techniques. Only AMPLICOR gave RNA titers above the cut-off value for all the patients. The techniques agreed satisfactorily, although there was a difference, median 0.4 log10, between the AMPLICOR and NASBA values. The non-progressors had low and stable RNA concentrations. The concentration was higher in the progressors, according to the AMPLICOR technique, from their inclusion in the study, and according to the NASBA technique, from 1 year after inclusion. However, only four of ten individual progressors had stable plasma HIV RNA concentrations significantly above those of the non-progressors before the decline in their CD4 counts. These were all and only the patients with a decline in lymphocyte counts more than 100 CD4/mm3/year. In each of the other progressors, the RNA concentration was not significantly different from those of the non-progressors. Thus, when making decisions about therapy, plasma HIV RNA determinations cannot be used in place of CD4 counts and may provide valuable additional information.