Background Malnutrition hampers educational performance of schoolchildren coming from low-income families. School feeding program was, thus, launched in public primary schools in Addis Ababa very recently. It is, thus, important to measure the initial nutritional status of participating students to see the effect of the program on their nutritional wellbeing.rnMethods The first-round survey was made at the initiation of the program. A multi-stage stratified sampling from 50 schools located in the ten sub cities of Addis Ababa yielded 4500 children of ages five to 19 years. Data was collected on age, height, weight and MUAC of the school children. The nutritional status of the children was evaluated using conventional anthropometric indicators, modified Composite Index of Anthropometric Failure (mCIAF), and MUAC-for-age. Receiver operating characteristic (ROC) regression was used to examine classification of malnourishment by MUAC-for-age versus BMI-for-age and mCIAF versus MUAC-for-age. Multilevel mixed effects model was applied to investigate variations in the prevalence of malnutrition across sub cities. Repeated measures ANOVA, was employed to assess differences in the outcomes over time.rnFindings The area under the ROC curves (AUC) for MUAC-for-age against BMI-for-age z-scores was 0.68 and that of mCIAF against MUAC-for-age was 0.70, respectively, indicating an overall better classification of malnourishment. Mixed effects model showed significant variations in nutritional status of schoolchildren across sub cities. Conventional measures showed that prevalence of stunting, thinness, or underweight among the sample children was 23.4%, 18.4%, and 16.5%, respectively. Assessment by mCIAF, instead, showed a higher prevalence of overall malnutrition (43.4%). MUAC-for-age indicated an acute malnutrition measurement of 33.4%. Significant differences (p