A study was conducted to determine the capabilities of the successive numerical Landsat data for assessment and monitoring of land use. Kashan plain with 7230 km2 of area, which is located in an arid zone of the central part of IRAN, selected as the site of investigation. It seemed to be a region prone to desertification processes. Two sorts of Landsat data: MSS (1976), TM (1998) and the supplementary information such as the soil and topography and land cover maps were collected.After preprocessing, the images were classified on the base of the field and subsidiary data. For Mss data, the Mss1, Mss2 and Mss3 were merged and showed the best correlation with field samples. In TM data, merging the TM3, Tm4 and Tm5 showed the best correlation. The classification performed by the minimum distance algorithm.Five land classes were distinguished with the overall precision of 65% and 75% for Mss and TM respectively. Detection of the changes between two maps showed a decrease of the area under range, bare and salt flats as much as 7.4%, 18.3% and 1.5% respectively, and an increase in cultivated and forest land use by the factor of 0.5% and 20.7% respectively. some referral to reliable documents determined the good agreement between the results and the real occurrence, except for the cultivated lands, that was significantly under estimated. It is concluded that thematic maps, which have released by image processing in a time series, could be compared and some general and useful results will be expected. Access to the same time field data of image acquisition time can promote the precision.