This work aims to redesign the large regions of Brazil based on agglomerations of municipalities with similar socioeconomic characteristics. Based on data from the last two censuses, 2000 and 2010, a matrix of 41 socioeconomic variables was selected for each one of the Brazilian municipalities. Then, the size of this matrix was reduced using the principal component analysis (PCA) method. The main components were selected to compose the algorithm of hierarchical agglomeration that defined the municipal clusters. The results of the agglomeration indicated the existence of eleven regions of municipalities for the year 2000 and ten regions for the year 2010, showing that the current political‐administrative map of the regions does not represent the junction of cities that maintain a relation of homogeneity.