OBJECTIVE: PI susceptibility results from a complex interplay between protease and Gag proteins, with Gag showing wide variation across HIV-1 subtypes. We explored the impact of pre-treatment susceptibility on the outcome of lopinavir/ritonavir monotherapy. METHODS: Treatment-naive individuals who experienced lopinavir/ritonavir monotherapy failure from the MONARK study were matched (by subtype, viral load and baseline CD4 count) with those who achieved virological response ('successes'). Successes were defined by viral load <400 copies/mL after week 24 and <50 copies/mL from week 48 to week 96. Full-length Gag-protease was amplified from patient samples for in vitro phenotypic susceptibility testing, with susceptibility expressed as fold change (FC) relative to a subtype B reference strain. RESULTS: Baseline lopinavir susceptibility was lower in viral failures compared with viral successes, but the differences were not statistically significant (median lopinavir susceptibility: 4.4 versus 8.5, respectively, P = 0.17). Among CRF02_AG/G patients, there was a significant difference in lopinavir susceptibility between the two groups (7.1 versus 10.4, P = 0.047), while in subtype B the difference was not significant (2.7 versus 3.4, P = 0.13). Subtype CRF02_AG/G viruses had a median lopinavir FC of 8.7 compared with 3.1 for subtype B (P = 0.001). CONCLUSIONS: We report an association between reduced PI susceptibility (using full-length Gag-protease sequences) at baseline and subsequent virological failure on lopinavir/ritonavir monotherapy in antiretroviral-naive patients harbouring subtype CRF02_AG/G viruses. We speculate that this may be important in the context of suboptimal adherence in determining viral failure.