Background: Severe acute malnutrition affects around 17 million under-five children in the world, of which the highest burden is accounted by Sub-Saharan Africa where Ethiopia is found. Though there are few individual, inconsistent and inconclusive studies, there is no nationally representative study on treatment outcomes of SAM in outpatient therapeutic feeding programs of Ethiopia. This study aimed at estimating the pooled treatment outcomes and predictors of recovery rate among under-five children with SAM in Ethiopia. Methods: Electronic databases (PubMed, Medline (EBSCOhost), EMBASE (Elsevier), CINAHL (EBSCOhost), web of science, Scopus, Science Direct and Food Science and Technology Abstracts (FSTA)), and grey literature sources (Google scholar, Mednar, World Cat and google) were used to retrieve articles. The random effect model was used to estimate the pooled treatment outcomes. Hazard ratios were used to determine the predictors of recovery rate. Cochran's Q, I 2 , and univariate Meta regression were done for heterogeneity. Begg's & Egger's tests were used for publication bias. Results: Nineteen articles with a total number of 23,395 under-five children with SAM were used for this metaanalysis. The pooled recovery, death, defaulter and non-recovery rates were 70% (95% CI: 64, 76), 2% (95% CI: 1, 2), 10% (95%CI: 7, 12), 15% (95% CI: 10, 20), respectively. Diarrhea (HR = 0.8, 95% CI: 0.75, 0.94), no edema (HR = 0.41, 95% CI: 0.33, 0.50) and amoxicillin (HR = 1.81, 95% CI: 1.18, 2.44) were independent predictors of recovery rate of children with SAM in Ethiopia. Publication year was found to be the potential source of heterogeneity between included studies.