The introduction of highly active antiretroviral therapy (HAART) has significantly improved life expectancy of HIV-infected patients; nevertheless, it does not eliminate the virus from hosts, so a cure for this infection is crucial. Some strategies have employed the induction of anti-HIV CD8 + T cells. However, the high genetic variability of HIV-1 represents the biggest obstacle for these strategies, since immune escape mutations within epitopes restricted by Human Leukocyte Antigen class I molecules (HLA-I) abrogate the antiviral activity of these cells. We used a bioinformatics pipeline for the determination of such mutations, based on selection pressure and docking/refinement analyses. Fifty HIV-1 infected patients were recruited; HLA-A and HLA-B alleles were typified using sequence-specific oligonucleotide approach, and viral RNA was extracted for the amplification of HIV-1 gag, which was bulk sequenced and aligned to perform selection pressure analysis, using Single Likelihood Ancestor Counting (SLAC) and Fast Unconstrained Bayesian Approximation (FUBAR) algorithms. Positively selected sites were mapped into HLA-I-specific epitopes, and both mutated and wild type epitopes were modelled using PEP-FOLD. Molecular docking and refinement assays were carried out using AutoDock Vina 4 and FlexPepDock. Five positively selected sites were found: S54 at HLA-A*02 GC9, T84 at HLA-A*02 SL9, S125 at HLA-B*35 HY9, S173 at HLA-A*02/B*57 KS12 and I223 at HLA-B*35 HA9. Although some mutations have been previously described as immune escape mutations, the majority of them have not been reported. Molecular docking/refinement analysis showed that one combination of mutations at GC9, one at SL9, and eight at HY9 epitopes could act as immune escape mutations. Moreover, HLA-A*02-positive patients harbouring mutations at KS12, and HLA-B*35-positive patients with mutations at HY9 have significantly higher plasma viral loads than patients lacking such mutations. Thus, HLA-A and -B alleles could be shaping the genetic diversity of HIV-1 through the selection of potential immune escape mutations.