Purpose The analysis of poverty is fundamentally focused on examining the well-being condition of the poor. We usually neglect the information provided by the rich. Nevertheless, perhaps the non-considered information indicating the determinants of non-poverty is also useful for fighting against poverty. The purpose of this paper is to analyze poverty under a new angle i.e. focusing on the information provided by the non-poor instead of the poor. For that a richness index is calculated in order to estimate econometric models regressing both indices i.e. poverty and richness indices on same selected characteristics. Thus, the comparison of the determinants of poverty and non-poverty for Tunisian case have allowed the classification of the selected explanatory variables with significant effect into four categories: the variables having significant effect on both sides (permanent effect), the variables having significant effect on the poor but not on the non-poor (transitory effect), the variables having significant effect on the non-poor but not on the poor (insurance effect) and the variables without any effect on both cases (neutral effect). This procedure is thus important given that it provides additional information and new way to enhance the targeting efficiency of the poor and fighting against poverty. Design/methodology/approach Using Tunisian data, an original procedure is proposed for calculating a richness index, defined based on the common formula of calculating the poverty index. Next econometric models are estimated regressing both the indices i.e. poverty and richness index on same selected characteristics. Findings The comparison of the determinants of poverty and non-poverty have allowed the classification of the selected explanatory variables with significant effect into four categories: the variables having significant effect on both sides (permanent effect), the variables having significant effect on the poor but not on the non-poor (transitory effect), the variables having significant effect on the non-poor but not on the poor (insurance effect) and the variables without any effect on both cases (neutral effect). Originality/value The analysis and the classification of the determinants of poverty according to the determinants of non-poverty is never made before in the litterature. This procedure is important given that it provides additional information and a new way to enhance the efficiency of targeting the poor and fighting against poverty.