Implementation of multilevel model is becoming a common analytic technique over a wide range of disciplines including social and economic sciences. In this paper, an attempt has been made to assess the application of multilevel logistic model for the purpose of identifying the effect of household characteristics on poverty status in Ethiopia using household income, consumption and expenditure (HICE) survey data of 2011. Households are classified as either poor or non-poor based on the absolute poverty line set at yearly per capita consumption of Birr 3781. Accordingly, the random intercept only model indicates the existence of differences in poverty status among households across regions. The result of random intercept and fixed slope model show that the rates of poverty for households residing in Afar, Somali, SNNP, Benishangul-Gumuz and Gambela regions were higher than the average of all regions, while the rates for households residing in Harari and Addis Ababa regions were low compared to the average of all regions. The random coefficient model showed that the random effects of place of residence vary across regions in explaining poverty status. Further, this model was more appropriate to explain the regional variation than a model with fixed coefficients or empty model with random effects. Thus, researchers should take the advantage of multilevel models to identify correlates of poverty when the data structure is hierarchical like HICE survey.