There are limited data of immunologic and virologic failure in Asian HIV-infected children using non-nucleoside reverse transcriptase inhibitor (NNRTI)-based highly activeantiretroviral therapy (HAART). We examined the incidence rate of immunologic failure (IF) and virologic failure (VF) and the accuracy of using IF to predict VF in Thai HIV-infected childrenusing first-line NNRTI-based HAART.Methods: Antiretroviral (ART)-naïve HIV-infected children from 2 prospective cohorts treated with NNRTI-based HAART during 2001-2008 were included. CD4 counts were performed every 12 weeks and plasma HIV-RNA measured every 24 weeks. Immune recovery was defined as CD4%≥25%. IF was defined as persistent decline of ≥5% in CD4%in children with CD4%<15% at baseline or decrease in CD4 count ≥30% from baseline. VF was defined as HIV-RNA>1,000 copies/ml after at least 24 weeks of HAART. Clinical and laboratory parameter changes were assessed using a paired t-test, and a time to event approach was used to assess predictors of VF. Sensitivity and specificity of IF were calculated against VF.Results: 107 ART-naive HIV-infected children were included, 52% female, % CDC clinical classification N:A:B:C 4:44:30:22%. Baseline data were median (IQR) age 6.2 (4.2-8.9) years, CD4% 7 (3-15), HIV-RNA 5.0 (4.9-5.5) log10copies/ml. Nevirapine (NVP) and efavirenz (EFV)-based HAART were started in 70% and 30%, respectively.At 96 weeks, none had progressed to a CDC clinical classification of AIDS and one had died from pneumonia. Overall, significant improvement of weight for age z-score (p = 0.014), height for age z-score, hemoglobin, and CD4 were seen (all p < 0.001). The median (IQR) CD4% at 96 weeks was 25 (18-30)%. Eighty-nine percent of children had immune recovery (CD4%≥25%) and 75% of children had HIV-RNA <1.7log10copies/ml.Thirty five (32.7%) children experienced VF within 96 weeks. Of these, 24 (68.6%) and 31 (88.6%) children had VF in the first 24 and 48 weeks respectively.Only 1 (0.9%) child experienced IF within 96 weeks and the sensitivity (95%CI) of IF to VF was 4 (0.1-20.4)% and specificity was 100 (93.9-100)%.Conclusion: Immunologic failure, as defined here, had low sensitivity compared to VF and should not be recommended to detect treatment failure. Plasma HIV-RNA should be performed twice, at weeks 24 and 48, to detect early treatment failure.Trial Registration: Clinicaltrials.gov identification number NCT00476606. © 2011 Bunupuradah et al; licensee BioMed Central Ltd.