The paper discusses a new methodology for the measurement of multidimensional poverty by applying it to rural areas in Brazil. Initially is proposed a relatively broad set of dichotomous indicators of insufficiencies related to the welfare of households and from those we tested the existence of latent dimensions using the Rasch model. It was noted that several sub-sets of indicators obey the restrictions of this model. This allows proving the existence of a multidimensional latent space for rural poverty in Brazil, as anticipated by theoretical concepts that attempt to overcome the limitations of measuring and identifying the poverty on a one-dimensional approach. We found that there is an extensive list of tests and procedures to verify the suitability of a selected set of indicators over a certain dimension, in order to meet the requirements of the model. This methodology can be an important alternative for the measurement and characterization of poverty and deprivation information for families. Despite being a work whose results are much aggregated in spatial terms, it can be an indication for an application to census data, to identify variations of multidimensional poverty at the municipal level. The importance and scope of the measurement of poverty from a multidimensional view is also strongly associated with its appeal in terms of an approach that goes in the direction of the determinants of poverty and its broader constitutive forms. This ensures that efforts toward this empirical approach are most useful for the formulation of policies for this area.