ABSTRACT Background: Malnutrition is associated both under nutrition and over nutrition which causes of the body gets un proper amount of nutrients to maintain tissues and organ function. Malnutrition and diet are the biggest risk factors for the global burden of disease. Objective: To determine associated factors to underweight among under-five year children in Ethiopia. Method: The data source for analysis was the 2016 EDHS data from under-five children weightto- age anthropometric index (Z-score) child underweight divided in to three categories: severely underweight (Z-score<-3.00), moderately underweight and normal ( ). In order to achieve our objective descriptive, cross-tabulation, standard ordinal logistic regression and multilevel ordinal logistic regression analysis were used. Thus several ordinal models such as POM, GOM, PPOM, ACM, CRM and SLM models were fitted. Result: The descriptive result revealed that about 7.08% of children were severely underweight and 16.72% were moderately underweight while 76.19% were normal. Among the fitted model, PPOM fits the data better than other models. The test of heterogeneity suggested that, the underweight varies among region and multilevel ordinal model fit data better than single level ordinal model. For selected multilevel PPOM shows that mother education, wealth index of family, religion of mother, birth order, birth type, sex of child, birth size of child, diarrhea two weeks before the survey, fever two weeks before the survey, mother BMI, age of child and duration of breast feeding were significant determinants of underweight among under-five children. Conclusion: In order to reduce severity of underweight, awareness creation efforts have to improve their economic status, improve the level of education of mothers. It is recommended that design and implement primary health care and nutrition programs which would fit the features of each region to safeguard children from nutritional deficiency. Key words: underweight, partial proportional odds model, multilevel partial proportional odds model, under-five year children, Ethiopia