Most of the antiretroviral (ARV) studies in Brazil have been reported in treatment-experienced and naive patients rather than in the setting of treatment interruption (TI). In this study, we analysed reasons given for TI and resistance mutations occurring in 150 HIV-1-infected patients who underwent TI. Of the patients analysed, 110 (73.3%) experienced TI following medical advice, while the remaining patients stopped antiretroviral therapy (ART) of their own accord. The main justifications for TI were: ARV-related toxicities (38.7%), good laboratory parameters (30%) and poor adherence (20%). DNA sequencing of the partial pol gene was successful in 137 (91.3%) patients, of whom 38 (27.7%) presented mutations conferring ARV resistance. A higher viral load prior to TI correlated with drug resistance (P < 0.05). Our results demonstrate that there are diverse rationales for TI and that detection of resistant strains during TI most likely indicates a fitter virus than the wild type. High viral loads coupled with unprotected sex in this group could increase the likelihood of transmission of drug-resistant virus. Thus, treating physicians should be alerted to this problem when the use of ARVs is interrupted.