OBJECTIVES: To evaluate the durability of three integrase strand transfer inhibitors (INSTIs) and two NRTIs in ART-naive individuals. METHODS: The study design was observational. Patients were HIV-positive, ART-naive subjects starting raltegravir, elvitegravir/cobicistat or dolutegravir with two NRTIs. The primary endpoint was time to treatment failure, i.e. occurrence of virological failure (first of two consecutive plasma HIV RNAs ≥200 copies/mL after 24 weeks) or INSTI discontinuation for any reason apart from simplification. Secondary endpoints were INSTI discontinuation due to toxicity/intolerance and CD4 count response. Survival analysis was done using Kaplan-Meier and Cox regression. RESULTS: Two thousand and sixteen patients were included: 310 (15.4%) started raltegravir-based regimens, 994 (49.3%) started dolutegravir-based regimens and 712 (35.3%) started elvitegravir/cobicistat-based regimens. Over a median of 11 months, 167 patients experienced treatment failure; the 1 year probability of treatment failure was 6.5% for raltegravir, 5.4% for dolutegravir and 6.7% for elvitegravir/cobicistat (P = 0.001). Sixty-eight patients (3.4%) discontinued INSTIs owing to toxicity/intolerance. By multivariable analysis, patients starting raltegravir had a 2.03-fold (95% CI = 1.2-3.2) higher risk and patients on elvitegravir/cobicistat a 1.88-fold (95% CI = 1.2-2.9) higher risk of treatment failure versus dolutegravir; there was no difference in risk of discontinuation due to toxicity/intolerance when comparing dolutegravir and raltegravir and marginal evidence for a difference when comparing elvitegravir/cobicistat and dolutegravir (adjusted relative hazard = 1.94 for elvitegravir/cobicistat versus dolutegravir, 95% CI = 1.00-3.76, P = 0.05). CONCLUSIONS: In our real-life setting, INSTI-based regimens showed high potency and durability. Among regimens currently recommended in Europe, those including dolutegravir are associated with a lower risk of treatment failure. © The Author(s) 2019. Published by Oxford University Press on behalf of the British Society for Antimicrobial Chemotherapy. All rights reserved. For permissions, please email: firstname.lastname@example.org.