The predictive value of virological response to lopinavir (LPV)/ritonavir (r) was assessed in 126 HIV-infected patients who failed antiretroviral therapy and had begun a rescue intervention based on LPV/r. At 3 months, subjects with < or =6 protease (PRO) resistance mutations showed a higher rate of virological response (HIV-RNA drop > 1 log or to <50 copies/ml) than patients with >6 PRO resistance mutations (77% versus 48%; p = 0.01). On the other hand, virological responders had greater mean LPV plasma trough levels than nonresponders (6.4 versus 3.9 microg/ml; p = 0.02). A positive correlation was found between LPV trough concentration and viral load reductions at 3 months under LPV/r (r = 0.23; p = 0.017). Overall, virological response was seen in 80.8% of patients with LPV trough levels >4.8 microg/ml while in only 52.5% of patients with lower LPV trough concentrations (p = 0.002). In the multivariate analysis, both < or =6 PRO resistance mutations and LPV trough levels >4.8 microg/ml were independent predictors of virological response to salvage therapy with LPV/r. A genotypic inhibitory quotient (GIQ) was estimated for each patient based on the ratio between LPV trough levels and the number of PRO resistance mutations. A positive strong correlation was found between GIQ and viral load reductions (r = 0.42; p = 0.002). Virological response was seen in 78% of patients with a GIQ >0.7 but only in 41.6% of those with lower GIQ (p = 0.004). When LPV trough levels >4.8 microg/ml, PRO resistance mutations < or =6, and GIQ >0.7 were all included in a stepwise multivariate analysis, GIQ remained as the main independent predictor of response to LPV/r.