There is evidence of errors of reconstruction of buildings in historical areas of cities. Due to the high cost of such mistakes choosing the best project from several options need to be well justified. This paper presents a multicriteria method of choosing the best object, called SNOD (scale of normalized and ordered differences). The method has two stages: a preliminary formal analysis of these alternatives is fully performed by the computer, and identifying the preferences of decision makers. The method is based on a verbal approach to the analysis of solutions. This takes into account the possibility of interdependence of criteria. To assess the effectiveness of the method was carried out a statistical simulation. This method has been applied for practical cases, reconstruction of buildings and the location of supermarkets in Vilnius. The article is partly supported by RFBR and program of Presidium of RAS MMIS.