In this study, we build what we call the Malagasy Multidimensional Poverty Index (MALAMPI), which is an augmented-MPI. Here, in addition to the standard MPI dimensions (health, education and living standards), we add an additional and highly important dimension, namely employment, which is generally the sole means of production owned by poor or deprived people. Another shortcoming of the MPI approach is that it does not enable gender inequalities analysis. This is surprising since two out of the three dimensions of the classical MPI are individual attributes. In this study, we also provide a new methodology aiming at computing gender sensitive MPI-type indicators. We use data from the 2012-2013 Malagasy MDGs national survey. Results show that adding the employment dimension to the MPI framework consistently increases Multidimensional poverty in Madagascar, the poverty headcount moving from 56% to 72%. Using our newly developed gender-sensitive method, we bring to light a significant gender gap (about 7% of increase at the expense of women), while the classical comparison of poverty level between female-headed households and male-headed households would have led to the conclusion that women are not disadvantaged. We also bring to light the fact that the gender gap does not necessary decrease when the household seems advantaged in terms of monetary living standards quintiles or in terms of household professional status.