BACKGROUND: Antiretroviral treatment (ART) coverage in Myanmar is well below average. This study describes retention and baseline predictors of prognosis from the largest ART program in the country. METHODS: A cohort analysis of adult patients who initiated ART during 2003-2007 was conducted, with follow-up until the end of 2009. The primary outcome was attrition [death plus losses to follow-up (LTF)]. Baseline variables were assessed as potential risk factors. The cumulative probabilities of death, LTF, and attrition up to 5 years were described using Kaplan-Meier estimates. Cox regression was used to calculate hazard ratios of attrition, overall and separately for 2 time periods on ART: 1-6 and 7-36 months. RESULTS: A total of 5963 adults enrolled in the program, providing 17,581 person-years of follow-up. Median age at baseline was 33 years [interquartile range (IQR): 28-38], 61% were men, 45% were in World Health Organization stage IV, and the median CD4 count was 71 cells per cubic millimeter (IQR: 29-164). There were 821 (13.8%) deaths and 389 (6.5%) LTF over the study period, with a 72% probability of being retained in care in the 5-year cohort. Double the rate of loss was contributed by death compared with LTF, and attrition was almost 4 times higher in the period 1-6 months compared with 7-36 months. In the multivariable analyses of the program overall, older age [adjusted hazard ratio (aHR): 1.56, 95% confidence interval (CI): 1.25 to 1.94], being male (aHR: 1.52, 95% CI: 1.25 to 1.85), World Health Organization stage IV (aHR: 1.44, 95% CI: 1.19 to 1.74), and body mass index <16 kg/m² (aHR: 2.13, 95% CI: 1.71 to 2.66) were independently predictive of attrition. CONCLUSIONS: The excellent retention over >6 years in this large cohort demonstrates that ART delivery at the primary care level in Myanmar is feasible and should encourage support for further ART expansion in the country.