Accurate assessment of plasma HIV RNA levels at low concentrations is clinically important. We evaluated a second-generation quantitative HIV RNA assay (NucliSens HIV-1 QT), and three simple adaptations of the NucliSens standard protocol to lower the lower cutoff level. The assays were evaluated in constructed panels with known HIV RNA concentrations and in clinical samples. Results were compared with those obtained with the first generation (NASBA HIV-1 QT) and with two other commercially available assays: the Amplicor HIV Monitor test and the Quantiplex assay. In a constructed panel, results obtained by NASBA QT were on average 0.13 log(10) copies/ml (SD 0.15) higher than those of NucliSens. The NucliSens assay could quantify HIV RNA in at least 50% of the samples down to 518 (2.71 log(10)) copies/ml and NASBA QT to 5.80 x 10(3) (3.76 log(10)) copies/ml). Both assays correlated well with the known input (R NucliSens = 0.99; R NASBA QT = 0.996), but results were more variable at lower input levels. With the three different ultrasensitive NucliSens adaptations, HIV RNA could be quantified in at least 50% of the samples down to 100 (2.00 log(10)), 46 (1.66 log(10)), and 10 (1.00 log(10)) copies/ml, respectively. In patient samples, Amplicor results were on average 0.11 (SD 0.20) log(10) copies/ml above, NucliSens 0.02 (SD 0.29) copies/ml above, and Quantiplex 0.13 (SD 0.19) copies/ml below the mean of the three assay results per sample. The variation remained the same over the range of RNA levels with all three assays. The NucliSens assay can quantify HIV RNA at lower levels than the NASBA QT and is comparable to other commercially available assays. The lower cutoff of the NucliSens can be lowered down to 10 copies/ml.