This paper tries to analyze, based on the micro data of the Pesquisa Nacional por Amostra de Domicílios (PNDA) – years 1992-1999 and 2001 – the share of the paid domestic service income between some types of poor rural families that presented negative growing taxes in the 90s, showing that in different cases, the paid domestic service income better explains the reduction of rural poverty than governmental transferences (retirements and pensions). We made a brief analysis of the evolution of different types of rural extensive families (and poor) in the 90s, besides a comparative analysis among the big regions of the country with respect to the average and per capita incomes of different types of rural families, adopting as reference 2001. We tried also to demonstrate that the reduction of the poor rural families is due to, in this case, because they are becoming non-occupied families, residents in the rural areas, and not only because of agricultural exodus (substitution of the agricultural activities for the non-agricultural activities) and/or rural-urban exodus. To test these hypotheses, we adopted as reference the rural poverty line created by Takagi et al (2001) – of U$ 1,08 – and the new tipology of extensive families developed by the Projeto Rurbano.