The aim of this study was to compare retention and re-entry to treatment between adolescent subjects treated with methadone, those treated with buprenorphine, and those treated with symptomatic (non-opioid) medication only. We used a retrospective file review of all patients aged less than 18 at first presentation for treatment for opioid dependence. The study was conducted at the Langton Centre, Sydney, Australia, an agency specialising in the treatment of alcohol and other drug dependency. Sixty-one adolescents (age range 14 - 17 years at the time of commencing treatment); mean reported age of initiation of heroin use was 14 +/- 1.3 years (range 11 - 16). Sixty-one per cent were female. The first episode of treatment was methadone maintenance in 20 subjects, buprenorphine in 25, symptomatic medication in 15; one patient underwent assessment only. These 61 subjects had a total of 112 episodes of treatment. Subjects treated with methadone had significantly longer retention in first treatment episode than subjects treated with buprenorphine (mean days 354 vs. 58, p<0.01 by Cox regression) and missed fewer days in the first month (mean 3 vs. 8 days, p<0.05 by ttest). Subsequent re-entry for further treatment occurred in 25% of subjects treated with methadone, 60% buprenorphine and 60% symptomatic medications. Time to re-entry after first episode of buprenorphine treatment was significantly shorter than after methadone treatment (p<0.05 by Kaplan - Meier test). Methadone maintenance appears to have been more effective than buprenorphine at preventing premature drop-out from treatment of adolescent heroin users.