Background: People recovering from heroin addiction need better treatments than are currently offered. The chronic relapsing nature of drug dependence means that helping a patient to achieve abstinence is often difficult. Naltrexone blocks the effects of ingested heroin; however, evidence is conflicting regarding the best delivery method. Objectives: The primary purpose of the trial was to evaluate the clinical effectiveness and cost-effectiveness of extended-release naltrexone versus standard oral naltrexone versus relapse prevention therapy without medication for opioid use disorder (OUD). Design: This was a 3-year, definitive, three-centre, three-arm, parallel group, placebo-controlled, double-blind, double-dummy, randomised controlled trial. Setting: Two specialist NHS outpatient addiction clinics: one in London and one in Birmingham. Participants: Planned study sample - 300 adult patients with OUD who had completed detoxification. Interventions: One iGen/Atral-Cipan Extended Release Naltrexone device (iGen/Atral-Cipan, Castanheira do Ribatejo, Portugal) (765 mg naltrexone or placebo) at day 0 of study week 1. Three weekly directly observed active or placebo oral naltrexone tablets (2 × 50 mg, Monday and Wednesday; 3 × 50 mg, Friday) at day 0 of study week 1 (for 4 weeks) and then an 8-week regimen of patient-administered dosing at the same dosing level. Main outcome measure: The primary outcome measure was the proportion of heroin-negative urine drug screen (UDS) results at the end of the 12-week post-randomisation time point. Results: Six patients were recruited and randomised to receive study interventions. Two patients had no positive UDS samples for heroin during the 12-week treatment period, one patient had only one positive UDS sample and the remaining patients had two, six and eight positive UDS results for heroin. All patients had at least one missed clinic visit (range 1-14). Conclusions: Considerable problems were encountered with (1) the stipulated requirement of a validated ‘detoxified’ status prior to the initiation of the study naltrexone, (2) the requirement for a consent coolingoff period and (3) delays awaiting the surgical implant procedure. Major upheaval to the organisation and delivery of NHS community treatment services across England led to extremely poor levels of actual entry of patients into the trial. Research-vital clinical and procedural requirements were, therefore, more challenging to implement. The potential therapeutic value of the opioid antagonist naltrexone still needs clear investigation, including comparison of the established oral form with the new ultra-long-acting depot implant formulations (for which no licensed products exist in Europe). Despite the small number of study participants, some tentative conclusions can be reached, relevant to potential future work. The blinding of the active/placebo medications appeared to be good. Self-report was not sufficient to detect instances of heroin use. Self-report plus UDS information provided a fuller picture. Instances of lapsed heroin use were not necessarily followed by full relapse, and future work should consider the lapse-relapse relationship. The prison release setting also warrants special consideration. In future, investigators should consider seeking ethics approval for studies in which clinical procedures to accelerate the treatment process are permitted, even if outside orthodox clinical practice, if they address a clinical need at the time of challenge and clinical risk. In addition, it may be appropriate to seek exemption from the ordinary requirement of a cooling-off period after securing consent because it is often essential to initiate treatment promptly.