Background: Antiretroviral therapy (ART) restores immune function and reduces HIV-related adverse outcomes. Patients on ART are recommended to periodic monitoring to ensure effective and durable treatment outcomes. Treatment failure is the main indication for ART switching and assessed by criteria used to define virologic, immunologic or clinical failure. Objective: The aim of the study was to determine incidence and clinical predictors of ART failure in ART user adult HIV patients in North West Ethiopia. Methods: A retrospective follow up study was conducted from June 1, to August 30, 2018 to determine the incidence and associated factors with first line ART failure among 315 adult ART users who started treatment between January 1, 2012 and December 31, 2017. Data regarding patients' socio-demographics, clinical characteristics, and treatmentrelated information were collected through review of their medical charts. Data were analyzed using SPSS version 21. Bi-variable and multivariable Cox proportional hazard model was used to identify predictors of treatment failure. P ≤ 0.05 was used to declare association. Result: Among 315 patients enrolled, 185 (59%) were females. Median age was 35 years (IQR 30-40 years). During 13,374 person-months of follow up, 32 (10.2%) persons failed their first line ART regimen at incidence of 2.39 failures per 1000 person-months of observation. Inadequate adherence (AHR (95% CI) = 6.01 (2.543-14.661), P value = 0.001), WHO clinical stage 3 and 4 (AHR (95% CI) = 4.471 (1.625-12.301), P value = 0.004), low base line BMI (AHR (95% CI) = 3.343 (1.552-7.201), P value = 0.002), and presence of ART drug toxicities (AHR (95% CI) = 2.470 (1.013-5.594), P value = 0.03) were found to be significant predictors of ART failure. Conclusion: Incidence of ART failure in ART clinic, University of Gondar hospital was considerable. Inadequate adherence was found to be the strongest predictor of first line ART failure. Close follow up and focused care is important to identify early treatment failure using the available resources.