The goal of antiretroviral therapy is reduction in morbidity and mortality via suppression of human immunodeficiency virus (HIV) viral load (VL) to undetectable levels. VL assay sensitivity has improved over time, but the reproducibility and clinical importance of VL results marginally higher than the limit of detection (LoD) are uncertain. We assessed the reproducibility and concordance of low VL results obtained with the Roche Cobas AmpliPrep/Cobas TaqMan HIV-1 version 2.0 (CAP-CTM) and the Abbott RealTime (m2000) HIV-1 assays, using longitudinal specimens from HIV-1-infected patients with low VL (<300 copies/ml) and stable CD4+ cell counts. Based on replicate testing of 3 specimens, coefficients of variation for log-transformed VL results were 5-8 % for m2000 and 9-10 % for CAP-CTM. The concordance between assays in specimens from patients with previously undetectable, detectable but not quantifiable VL, or variable (undetectable/detectable but not quantifiable VL) results over time was 90, 56, and 56 %, respectively. Correlation between results for specimens with quantifiable VL (initially 40-300 copies/ml) was moderate (R (2) = 0.48) with significantly higher results for CAP-CTM and a mean difference (CAP-CTM minus m2000) of 0.10 log(10) copies/ml. T-cell activation (CD8+/CD38+ percentage) in patients with low VL was initially higher than in patients with undetectable VL, and then decreased to equivalent levels over time. These results indicate that residual viremia at levels slightly above the LoD have no negative effect on T-cell activation.