ABSTRACT: BACKGROUND: Long-term outcomes of antiretroviral therapy (ART) in children remain poorly documented in resource-limited settings. The objective of this study was to assess two-and three-year survival, CD4 evolution and virological response among children on ART in a programmatic setting in Cambodia. METHODS: Children treated with first-line ART for at least 24 months were assessed with viral load testing and genotyping. We used Kaplan-Meier analysis for survival and Cox regression to identify risk factors associated with treatment failure. RESULTS: Of 1168 registered HIV-positive children, 670 (57%) started ART between January 2003 and December 2007. Survival probability was 0.93 (95% CI: 0.91-0.95) and 0.91 (95% CI: 0.88-0.93) at 24 and 36 months after ART initiation, respectively. Median CD4 gain for children aged over five years was 704 cells/mm3 at 24 months and 737 at 36 months. Median CD4 percentage gain for children under five years old was 15.2% at 24 months and 15% at 36 months. One hundred and thirty children completed at least 24 months of ART, and 138 completed 36 months: 128 out of 268 (48%) were female. Median age at ART initiation was six years.Overall, 22 children had viral loads of >1000 copies/ml (success ratio = 86% on intention-to-treat-analysis) and 21 of 21 presented mutations conferring resistance mostly to lamivudine and non-nucleoside reverse transcriptase inhibitors. Risk factors for failure after 24 and 36 months were CD4 counts below the threshold for severe immunosupression at those months respectively. Only two out of 22 children with viral loads of >1000 copies/ml met the World Health Organization immunological criteria for failure (sensitivity = 0.1). CONCLUSIONS: Good survival, immunological restoration and viral suppression can be sustained after two to three years of ART among children in resource-constrained settings. Increased access to routine virological measurements is needed for timely diagnosis of treatment failure.