The objective of the thesis is to inquire into the robustness of poverty lines and povertyrnindices. This research was undertaken to contribute to the work on poverty analysis by usingrnthe household survey data, to investigate to an area not researched before. The conclusionrnof the paper leads to policy recommendations.rnrnThe data used is drawn from the Ethiopian Household Survey conducted for 15 rural sites andrnseven urban sites in 1994. The methodologies used include the calculation of four types ofrnpoverty indices by using two types of poverty lines constructed by the following methods i. ernCost of Basic Needs (CBN) and the Food Energy Intake (FEI), the FGT measures (i.e thernHead-count index, the Poverty-gap Index and P2), SEN's index, Thons' index and CHU'srnindex. The results obtained for both the rural and urban sites were further analyzed to seerntheir robustness by cross tabulation method; by Persons rank correlation; regression methodrnand by calculation of correlation coefficients.rnrnIt was found that generally the CBN and FEI do not identify exactly the same households asrnpoor. In the case of the head-count index, CBN and FEI are not robust for urban sites, theyrndo not give the same number of poor, but for rural sites they are robust. The results of thernvarious indices tested by different methods lead to the conclusion that poverty indices arernsensitive to the type of poverty line used, especially SEN's index. Even using the samernpoverty line, all the indices do not give the same ranking in the majority of the sites. In thernanalysis of the correlates of poverty, both poverty lines give very close results thus they arernrobust in this case.rnThe policy recommendations from this thesis are that the choice of a particular poverty linernis important for targeting the poor