The aim of this study is to test the feasibility of the detection and mapping of annual clearcuts from Landsat Thematic Mapper data over the forested region les Landes, in the south-west of France ( about 1 million hectares covered by uniform age monospecific forest stands of maritime pine) Two Landsat TM images acquired in summer 1985 and 1986 were used. The radiometric changes following a clearcut are analysed and compared to the changes following up other types of evolution, on a test area representing 200000ha. Newly clear felled stands showed strong increase in their mean radiometric level in the visible as well as in the middle infrared, but no clear variation in the near-infrared was found. Results of supervised image classifications using three different band combinations were compared to evaluate the potential contribution of the nearinfrared combined with the middle infrared for the detection of annual clear fellings. The following multi-temporal combinations were used: (TM–5–85, TM5– 86), (TM–5–85, TM–5–86, TM–4–86), (TM–5–85, TM–4–85, TM–5–86, TM–4–86). The combinations using the middle infrared alone led to the best identification of new clearcuts harvested between 1985 and 1986, but the commissionerror was high (about 25 per cent). The main confusion was observed on old clearcuts which were reforested between 1985 and 1986. In order to propose a global method for mapping and estimating the annual clearcuts over the whole Landes region, it is necessary to test whether the results are reproducible when images acquired at very different dates are used. Moreover, the cartographic accuracy must be controlled and improved. Effects of thinnings, which are not studied here, must also be taken into account. Then, we can propose a global system for monitoring regional forest resources.