Poverty is necessarily a multidimensional concept. The need to rank countries, regions, neighbohoods, time periods, social groups even families requires, however, a scalar representation. A traditional solution has been to limit the concept of poverty to insufficient income. Recently, however, great emphasis has been given to the construction of scalar measures that take explicitly into consideration the multidimensional nature of poverty. However, several of theses indicators, as the Human Poverty Index (HPI) developed by United Nations Development Program (UNDP), share a severe inconvinience: They are not suited to provide estimates of the degree of poverty for each family, since they were originally designed to obtain estimates manily for countries and regions. In this study we aim to overcome this limitation, by introducing a scalar indicator specifically designed for estimating the degree of multidimensional poverty of each family from commonly available household surveys, like PNAD. This indicator was constructed to be additively separable as proposed by Chakravarty, Mukherjee and Ranade (1998).