Methadone maintenance treatment (MMT) in China was established in 2004. The purpose of the present study was to estimate client retention and identify client factors associated with longer times in treatment.Data were abstracted from the MMT Data System from April 2008 to March 2010. Clients were considered to have interrupted treatment when they missed 30+ days of treatment. The median time in treatment was estimated using Kaplan–Meier methods and factors associated with the duration of treatment were evaluated using accelerated failure time models.Among 107,740 clients enrolled in MMT between 2008 and 2010, the median time spent in MMT among clients was 155 days but the majority (69%) did not attend every day. Estimated probabilities for treatment interruption were 53% at 6 months, 66% at 12 months and 77% at 24 months. Longer time in treatment was associated with doses ≥60mg (Time ratio (TR)=2.12, 95%CI=1.96–2.30) and having tested negative on their last urine opiate test (TR=2.15, 95%CI=2.03–2.27). The effect of continued drug use was significantly modified by attendance. Sensitivity analyses indicated a dose-response relationship.Irregular attendance suggests there are barriers to accessing services that need further attention. The Chinese MMT programme needs effective strategies to improve its treatment durations.