Background: Virological failures on combined antiretroviral therapy still occur. Boosted protease inhibitor (PI/r)-based therapy is a commonly used option after non-nucleoside reverse transcriptase inhibitor (NNRTI) failure, but whether two fully active nucleoside reverse transcriptase inhibitors (NRTIs) are required is unknown. We investigated the effect of an NRTI backbone in individuals receiving PI/r after failing NNRTI-based combined antiretroviral therapy. Methods: A longitudinal analysis of the UK Collaborative HIV Cohort (CHIC) and UK HIV Drug Resistance Database to identify individuals who failed irst-line NNRTI and two NRTIs, and switched to PI/r-based therapy between January 1999 and December 2008 was conducted. We investigated the effect of NRTI on suppression. Results: In total, 470 individuals met study criteria: 19.6%, 34.5% and 46.0% started 0, 1 or =2 NRTIs, respectively. Median CD4+ T-cell count was 223 cells/mm3 and HIV-RNA was 4.3 log10 copies/ml; 246 (52.3%) underwent genotyping before switch. Virological failure occurred in 10.9% and 13% after 48 and 96 weeks, respectively. In multivariable analysis, heterosexual risk group and HIV RNA were independently associated with virological failure; higher CD4+ T-cell count was protective (HR=0.92). Number of new NRTIs or genotypic sensitivity score of backbone had no effect on treatment success rates when modelled as categorical or continuous variables. Conclusions: Successful treatment with a second-line PI/r may not require two active NRTIs. If replicated in clinical trials, these indings could guide future recommendations.