t Some authors argue that deforestation in Africa is one of the highest among the world's continents. However, for several regions, historical data and observations obtained by remote sensing raise some controversy over the portrait accepted in international circles. This case study, located in the central plateau of Angola, highlights the importance of medium to high resolution multi-temporal observations of land cover by remote sensing as a means of obtaining objective information and illustrates the complexity of deforestation processes. Land cover change in the Huambo province (Angola) is analyzed using maps produced for three dates (in the period between 1990 and 2009) by classification of Landsat TM and ETMþ images. The results are a first stepping stone for the development of a forest monitoring baseline for REDD projects under the scope of post-Kioto agreements, and they indicate an overall recovery in the forest extent of Huambo (negative deforestation rate of À0.16%, contrasting with the þ0.20% reported for the entire country). However, this apparent recovery hides an inversion in the dynamics of the more densely wooded vegetation types, the miombo woodlands, which show a very slight expansion in the first decade analyzed and a marked reduction in the second, while agriculture shows a consistent and constant expansion. The deforestation rate of 1.49% in miombo woodlands in the last decade is found to be seven times higher when compared with official figures reported for the entire country. These results are discussed in the context of the socio-economic factors that may have driven the observed dynamics.