In resource-limited countries (RLCs), WHO recommends HIV viral load (VL) on dried blood spots (DBS) for antiretroviral therapy (ART) monitoring of patients living in non-urban settings where plasma VL is not available. In order to reduce the impact of proviral DNA interference, leading to false positive results in samples with low plasma VL, we compared three different nucleic acid preparation methods with the NucliSens (Biomérieux) extraction, known for its high recovery of nucleic acids on DBS. Paired plasma-DBS samples (n = 151) with predominantly low plasma VL (≤ 10,000 copies/ml; 74%) were used. At the threshold of 1,000 copies/ml on DBS, 51% and 10% were misclassified as false positives or false negatives, respectively with NucliSens, versus 41% and 20% with m2000sp (Abbott), described as more specific for RNA recovery. DNase treatments of nucleic acid extracts and free virus elution (FVE) protocol before nucleic acid extraction, reduced the proportion of false positives to 0% and 19%, but increased the proportion of false negatives to 40% and 73%. More efforts are thus still needed to improve performance of VL assays on DBS to monitor patients on ART in RLCs and allow timely switch to more costly second or third line ART regimes.