The choice to include the optimal protease inhibitor (PI) in highly active antiretroviral therapy (HAART) regimens in children with perinatal HIV-1 infection is still under debate. Virologic and immunologic outcomes of three different regimens in an observational paediatric cohort were compared. Data from 12 saquinavir-, 18 nelfinavir-, and 10 lopinavir/ritonavir-treated children were analyzed after 4 and 24 weeks of therapy. Immunologic and virologic outcomes were compared using multivariate analysis adjusting the results for age, baseline CD4+ T-lymphocyte count and baseline viral load. Saquinavir-treated children displayed significant reduction in viral load at week 24 (but not at week 4) and no increase in CD4+ T-lymphocyte count, indicating a poor advantage in using this drug. Lopinavir/ritonavir-treated children presented lower viral loads than nelfinavir-treated children at week 4 (P=0.020) and week 24 (P<0.0001). Virologic failure occurred in 6/18 (33.3%) nelfinavir-treated children but in no child receiving lopinavir/ritonavir (P=0.013). An undetectable viral load was achieved in 9/10 (90.0%) lopinavir/ritonavir- vs. 3/18 (16.6%) nelfinavir-treated children (P<0.0001). No significant difference in CD4+ T-lymphocyte count was observed between lopinavir/ritonavir- and nelfinavir-treated children at weeks 4 and 24. However, a different kinetic of the immunologic recovery was observed. Lopinavir/ritonavir-treated children displayed higher CD4+ T-lymphocyte counts than saquinavir-treated children since the first month of therapy (week 4: P=0.042; week 24: P= 0.029) while nelfinavir-treated children took 24 weeks to reach such an outcome (P=0.034). Since lopinavir/ritonavir-based regimen controls viral replication more efficiently and restores CD4+ T-lymphocyte count more quickly than saquinavir- or nelfinavir-based HAART, it may be considered when a salvage therapy or a rapid increase in CD4+ T-lymphocytes is necessary.