OBJECTIVES: To study factors associated with risk of dropout among heroin addicts enrolled in a community-based methadone maintenance treatment (MMT) clinic in Solo, Central Java, Indonesia. METHODS: This was an ambi-directional cohort study. The index date was patients' first entry identified from medical records. Patients enrolled from September 2009 were followed up until departure from the clinic or otherwise censored in May 2011. Other data collected were methadone doses, risk behaviour, and sociodemographic characteristics. Kaplan-Meier method was used to estimate retention rate and Cox proportional hazard regressions were used to determine factors associated with dropout from treatment. RESULTS: Ninety-eight patients aged 31.6 years old on average contributed to 14,804 person-days of folllow up. The retention rate was 24.2% after 21 months with median retention time of 78 days. The median and maximum dose were 43.1 mg and 150 mg, respectively. Multivariate analysis showed that methadone dose below 80 mg (HR=2.4, p=0.014, 95% CI=1.2-4.8) and absence of family support (HR=5.0, p<0.001, 95% CI=2.4-10.6) were significantly associated with risk of dropout. HIV status, overdose history, criminal record, duration of addiction, distance from clinic, marital status, and employment were not statistically significant predictors of dropout. CONCLUSIONS: Despite high prevalence of injecting drug users (IDUs) in the area, low retention rate of the MMT clinic indicates that the community-based program might have been suboptimally utilised or else poorly managed. Attaining maintenance phase at a minimum dose of 60 mg was insufficient to ensure compliance. Health workers should maintain relationships with patients' family to help support and monitor the treatment. Similar studies in hospital or correctional institution settings and/or other areas are needed to confirm the findings as well as economic analyses to anticipate waste of resources.