The repopulation of abandoned areas in Angola after the 27 years of civil war led to a fast and extensive expansion of agricultural fields to meet the rising food demand. This effect is supported by the improvement of infrastructure and has a large impact on the natural vegetation.We evaluated agricultural expansion in our study area in the south of Angola by using Landsat data with a resolution of 30 x 30 m per pixel, combined with RapidEye data (5 x 5 m per pixel) and high resolution imagery (Bing Maps, 0.41 x 0.41 m and Google Earth, 1.5 x 1.5 m). We applied a bi-temporal change detection approach on a spectral difference Landsat image, based on four images for 1997/1998 and 2008/2009 each. After calculating an unsupervised classification (ISODATA), change classes were manually assigned based on spectral information from Landsat data and visual information of the higher resolution data. In order to analyse which land cover type was most affected by the agricultural practices, we additionally carried out an unsupervised classification approach in combination with knowledge-based assignment of classes for 1997/1998. We furthermore estimated the relationship between infrastructure and the establishment of new fields by manually capturing roads and tracks based on high resolution imagery (Bing Maps & Google Earth) and recent RapidEye data.The results indicate that together with the repopulation of the study area, the area of fields strongly increased, which had a considerable impact on natural vegetation. Furthermore, the high impact of the presence and quality of infrastructure on the location and the spatial expansion of fields could be proved. We found that more than 47,000 ha (equals 2.85 % of the total study area) of new fields were established during the 11 years and mainly open woodlands, shrublands and dense woodlands were cleared for slash and burn agriculture. Furthermore, more than 93% of these active agricultural areas are located within less than 5 km distance to the roads and 46% even lie within a 1-km distance to the present infrastructure. Keywords: Change analysis, deforestation, ISODAT, Landsat 5 TM, land use classification, Okavango, slash-and-burn.