Apart from the BENCHMRK study, there are no large observational experiences describing the long-term efficacy and safety of rescue regimens for human immunodeficiency virus type 1 (HIV-1) infection. Antiretroviral-experienced patients with detectable viraemia starting a raltegravir (RAL)-based regimen between March 2007 and June 2009 were consecutively enrolled and followed for ≥4 years. Data were censored at Week 206 for homogeneity. Of 333 patients, 258 (77.5%) were still on RAL-based therapy at Week 206, and 241 had undetectable HIV-1 RNA (73% in intention-to-treat analysis). Of the 75 subjects who discontinued RAL therapy, 36 were lost to follow-up, 15 changed their regimen due to virological failure, 2 simplified their regimen stopping RAL, 9 stopped all antiretrovirals and 13 died. Overall, 100 subjects (30.0%) had at least one detectable viraemia, but only 32 (9.6%) had true viral failure. Seventeen patients continued their failing regimen. ‘Blips’ were experienced by 53 patients (15.9%), whilst 15 (4.5%) had confirmed viral rebound due to adherence issues and were re-suppressed upon treatment re-introduction. In a multivariate analysis of predictors of interruption or failure, each baseline HIV-1 RNA log10 increase was associated with an adjusted hazard ratio for failure of 1.6; having more than 13 previous treatment courses also emerged as a predictor. Overall, adverse events were rare (n=64), with 13 deaths. Tumours were mainly early events, often fatal (7/15), mainly non-Hodgkin's lymphomas (8), followed by hepatocarcinoma (2). RAL proved effective and well tolerated in this cohort, and few patients experienced viral failure after 4 years.