Landsat satellite images of Landsat are used for land observation, they have a long-term large archive and are freely available, therefore they are especially usefull for time-series. In this thesis we carried out a time-series of built-up areas in Ljubljana and Houston, based on Landsat images acquired between the years of 1987 and 2017. We used the supervised classification with a maximum likelihood classification algorithm, whose results are six classifications for each area. Based on one hundred random points, we evaluated each classification. By comparing classifications we made change maps, which show changes in the built class, according to other classes (forest, water and agricultural land). An analysis of the of enlargement built-up areas is presented. A comparison of the study areas is described.