The evaluation of real estate is usually made using Classic Model of Regression. However, how the data are distributed in space an additional factor must be considered: the spatial autocorrelation. The non consideration of this factor can generate serious problems to the work. In this case, the methodology traditionally adopted it is inadequate and it should be substituted by a new methodology denominated Space Regression. This work shows that metodology and analyzes 59 data market of the city of Recife-PE. The obtained results prove the existence of spatial dependence in the data.