The aim of this study was to establish the prevalence and predictors of genotypic resistance of HIV-1 to lopinavir and fosamprenavir from patients failing protease inhibitors (PI)-based regimens. We selected 643 HIV-1-infected patients with available treatment history who underwent genotypic resistance assays for virological failure from a clinical site and from the Stanford database. According to the genotypic resistance interpretation of the Stanford algorithm, proportions of viruses showing full susceptibility to fosamprenavir and lopinavir were 32% and 34%, respectively (p =ns). Proportions of viruses fully susceptible to lopinavir/r and fosamprenavir/r according to the Agence Nationale pour la Recherche sur le SIDA (ANRS) algorithm, were 81% and 81%, respectively. According to the Rega algorithm, proportions of viruses showing full susceptibility to fosamprenavir/r and lopinavir were 80% and 70%, respectively (p<0.001). According to the ANRS and Rega interpretations, the time on therapy predicted susceptibility to lopinavir/r, while susceptibility to fosamprenavir/r according to ANRS was predicted by the number of prior PI regimens experienced. According to the Stanford interpretation, prior indinavir exposure predicted resistance to lopinavir/r and fosamprenavir/r while prior nelfinavir use predicted susceptibility to both drugs. After failing PI-based regimens, the majority of viruses retained a predicted susceptibility to fosamprenavir/r and lopinavir/r. In patients failing PIs, the interpretation of genotypic resistance to fosamprenavir may change considerably according to the different algorithms and in respect to the effect of pharmacokinetic boosting with ritonavir.