Factors predicting 2-year retention in methadone maintenance and the impact of repeat treatment episodes on retention are examined. Data (n=9555 episodes) were drawn from a population-based treatment registry and analyses were performed using episode-weighted data. We estimated a binary logistic regression model with a duration of 730 days or greater as the dependent variable. The odds of remaining in treatment for 730 days or more increase with age and vary by region and provider type, but decrease with increasing number of treatment episodes. In comparison with other studies, these analyses show much higher rates of retention in methadone treatment but suggest that repeat episodes may not be as beneficial as existing research suggests.