In this study, named the Zephir study (Telzir-pharmacokinetics), 121 antiretroviral-experienced human immunodeficiency virus (HIV) patients failing on highly active antiretroviral therapy (HAART) were included in a prospective cohort and received a fosamprenavir-ritonavir (700 mg/100 mg twice a day)-based regimen. The impact of baseline HIV type 1 (HIV-1) mutations, pharmacokinetic (PK) parameters, and genotype inhibitory quotient (GIQ) on the virological response at week 12 (W12) was assessed. HIV reverse transcriptase and protease were sequenced at W0. The response at W12 was defined as<2.3 log10 HIV-1 RNA copies/ml or a virus load decrease of>or=1 log10 copies/ml. W4 amprenavir PK were determined by high-performance liquid chromatography. Patients had a median of nine previous treatments over 8 years. Median W0 values were as follows: 295 CD4+/microl, 4.4 log10 HIV-1 RNA copies/ml, and 6 protease- and 5 nucleotide reverse transcription inhibitor-related mutations. Respective values for minimum concentration of drug in serum (Cmin) and area under the concentration-time curve (AUC) from 0 to 24 h were 1,400 ng/ml and 35 mg.h/ml. At W12, 52% of the patients were successes, with a median decrease of -0.7 log10 HIV-1 RNA copies/ml. The Zephir mutation score included 12 IAS protease mutations associated with poorer virological response: L10I/F/R/V, L33F, M36I, M46I/L, I54L/M/T/V, I62V, L63P, A71I/L/V/T, G73A/C/F/T, V82A/F/S/T, I84V, L90M, and polymorphism mutations I13V, L19I, K55R, and L89M. Comparing<4 versus>or=4 mutations, HIV-1 RNA decreases were -2.3 log10 copies/ml versus -0.1 log10 copies/ml (P<10(-4)) with 93% versus 19% successes (P<10(-4)), respectively. This score predicted W12 failure with 94% sensitivity, versus 31% for the ANRS 2005 algorithm. Cmin (<1,600 ng/ml), AUC (<40 mg.h/ml), and GIQ (<300) values were associated with failure (all P values were <10(-4)). The need to test genotype-based algorithms using different patient databases before their implementation in clinical practice is highlighted. Specific mutations, PK and GIQ, provide relevant information for monitoring fosamprenavir-ritonavir-based HAART.