Data warehousing and On Line Analytical Processing (OLAP) are technologies intended to support business intelligence. Spatial OLAP integrates spatial data into OLAP systems. Spatial OLAP models reformulate main OLAP concepts to define spatial dimensions and measures, and spatio-multidimensional navigation operators. Spatial OLAP reduces geographic information to its spatial component without taking into account map generalization relationships into the multidimensional decision process. In this paper, we present the concept of Geographic Dimension which extends the classical definition of spatial dimension by introducing map generalization hierarchies, as they enhance analysis capabilities of SOLAP models and systems. A Geographic Dimension is described by spatial, descriptive and/ormap generalization hierarchies. These hierarchies permit to define ad-hoc aggregation functions, but at the same time raise several modeling problems.