BackgroundrnMonitoring patient response with viral load, which is the gold standard, is not feasible inrnresource limited settings. Therefore it’s essential to look for factors that can be used tornidentify those patients at higher risk of treatment failure in these settings.rnObjectivernTo explore factors that can be used to identify those at a higher risk of treatment failure.rnMethodrnA Nested case-control study from a cohort of HIV patients on ART at governmentrnhospitals in Addis Ababa was conducted through review of medical records. A totalrnsample size of 423 with 141 patients with failure of their first line regimen (cases) andrn282 patients without failure (controls) is used. Base line socio-demographic and clinicalrninformation were collected. Comparison of survival times were made through KaplanrnMeier and Log-rank tests. Independent predictors of treatment failure were identifiesrnusing multivariate COX regression analysis.rnResultsrnThe mean survival time (without treatment failure) was 53 months (95% CI, 50 – 57).rnFemales were found to have a higher survival time of 57months (95%CI, 52-62, P= 0.01)rnand males have a significantly higher risk of developing treatment failure with anrnadjusted HR of 1.518 (95%CI, 1.084-2.125, P=0.01). Those with two or more episodes ofrnpoor adherence during their follow-up have a significantly higher adjusted hazard ratio ofrn4.02 (95% CI, 2.71, 5.96, P=