Remotely sensed data and related processing techniques have proven an effective tool for urban areas study, so that the integration of such products with other cartographic and thematic mapping products can help the management of the urban environment and ease the work of policy makers in city planning and administration. The focus of this work is on the topic of urban areas mapping over province administrative unit of Milan, Italy, using mid-resolution remote sensing data covering the last 20 years. In detail, for years 2008 and 1998 SPOT4 HRVIR data were used, while for 1989 Landsta5 TM data were used. The methodology applied consisted of three main steps; the first step exploits a pixel-based binary cascade thresholding of spectral bands and spectral indexes, which resulted in a 10 classes land cover map comprising both urban and non urban features. The second step filters the pixel-based urban maps using an object-base approach to map agricultural cover features as vegetated and bare fields, better distinguished as objects for their spatial texture and homogeneity. The third step involves the joining of land cover classes into two main classes: urbanized and non urbanized surfaces, and a final polishing of results made with expert knowledge visual interpretation of both satellite images and ancillary vector data. Final derived urban maps have been validated and demonstrated to provide good accuracy, with Overall Accuracy higher than 87% for automatically derived products and higher than 93% for expert refined products.