ABSTRACT: BACKGROUND:Untreated opioid addiction in people with HIV is associated with poor HIV treatment outcomes. Slow-release, long-acting, implantable naltrexone might improve these outcomes. Here, we present results of a study aimed to test this hypothesis. METHODS:We did a 48 week double-blind, double-dummy, placebo-controlled, phase 3, randomised trial with men and women addicted to opioids who were starting antiretroviral therapy (ART) for HIV and whose viral loads were higher than 1000 copies per mL. Participants were seeking treatment at two HIV and two narcology programme centres in Saint Petersburg, Russia, and the surrounding Leningrad region. The Pavlov statistical department created a table with stratification on gender distribution, viral load, and CD4 cell count. We stratified participants according to gender, viral load, and CD4 cells per ?L, and randomly assigned (1:1) them to addiction treatment with a naltrexone implant and oral naltrexone placebo (implant group) or oral naltrexone and placebo implant (oral group). The primary outcome was plasma viral load of less than 400 copies per mL at 24 weeks and 48 weeks. We included all randomly assigned participants in outcome analyses (intention to treat). Treatment staff and patients were masked to group assignment. The study is complete and registered at ClinicalTrials.gov, NCT01101815. FINDINGS:Between July 14, 2011, and April 14, 2014, 238 potential participants were recruited and screened, 35 were excluded for not meeting inclusion criteria, three declined to participate, and 200 were randomly assigned to treatment (100 to each group). At week 24, 38 (38) participants in the implant group and 35 (35%) in the oral group had viral loads less than 400 copies per mL (risk ratio 1·1, 95% CI 0·76-1·56; p=0·77). At week 48, 66 participants in the implant group and 50 in the oral group had viral loads less than 400 copies per mL (risk ratio 1·32, 95% CI 1·04-1·68; p=0·045). There were seven serious adverse events: three deaths in the implant group (one due to heart disease, one trauma, and one AIDS), and four in the oral group (two overdoses, one pancreatic cancer, and one AIDS). The overdose deaths occurred 9-10 months after the last naltrexone dose. INTERPRETATION:The longer the blockade of opioid effects, the more protection an individual gets from missed ART doses and impulsive behaviours that lead to relapse and poor, even fatal, outcomes. Commercial development of implants could result in a meaningful addition to addiction treatment options. FUNDING:National Institutes of Health, National Institute on Drug Abuse, Penn Centre for AIDS Research, and Penn Mental Health AIDS Research Centre.