ANRS 127 was a randomized pilot trial involving naïve patients receiving two dual-boosted protease inhibitor (PI) combinations. Virological response, defined as a plasma HIV RNA level of <50 copies/ml at week 16, occurred in only 41% patients. Low baseline plasma HIV RNA level was the only significant predictor of virological response. The purpose of this study was to investigate the impact on virological response of pretherapy mutations in cleavage sites of gag, gag-pol, and the gag-pol frameshift region. The whole gag gene and protease-coding region were amplified and sequenced at baseline and at week 16 for 48 patients still on the allocated regimen at week 16. No major PI resistance-associated mutations were detected either at baseline or in the 26 patients who did not achieve virological response at week 16. Baseline cleavage site substitutions in the product of the gag open reading frame at positions 128 (p17/p24) (P = 0.04) and 449 (p1/p6(gag)) (P = 0.01) were significantly more frequent in those patients not achieving virological response. Conversely, baseline cleavage site mutation at position 437 (TFP/p6(pol)) was associated with virological response (P = 0.04). In multivariate analysis adjusted for baseline viral load, these 3 substitutions remained independently associated with virological response. We demonstrated here, in vivo, an impact of baseline polymorphic gag mutations on virological response in naïve patients receiving a combination of two protease inhibitors. However, it was not possible to link the substitutions selected under PI selective pressure with virological failure.