The replicative capacity of HIV is studied by carrying out replication-competition experiments with the insertion of the gene of interest. These assays cannot capture the complicated patterns of mutations of different genes.A cross sectional study was carried out on 10 HIV-infected nai;ve patients and on 15 patients failing HAART. The CD8-depleted PBMCs, with known proviral DNA and cellular HIV-RNA copy numbers, were cultured. A reference curve was determined using the data obtained from 10 nai;ve patients. The replicative capacity was calculated as the ratio multiplied by 100 of the p24 antigen level of isolates over the p24 antigen level determined on the reference curve.A linear correlation between p24 antigen level and the infectious doses of HIV-DNA alone or plus cellular RNA copy number of PBMCs was found in naive patients (r=0.63, P<0.001 and r=0.67, P<0.001, respectively). Although all patients failing therapy had strains with impaired replicative capacity, a wide range of values (0.1-74.5%) was detected. All strains with a replicative capacity above 10% had non-nucleoside reverse transcriptase inhibitors related mutations.A viral assay to evaluate the HIV replicative capacity is described. The high variability of replicative capacity confirms the need to undertake replicative capacity assay using the whole virus.