There is an increasing interest on area-wide and high-resolution data of forest composition. In Switzerland, tree species distribution will be considered periodically by the Swiss National Forest Inventory (NFI), but the claims will be only partly fulfilled by the existing forest type maps since they are relatively poor regarding spatial accuracy, updating, and reproducibility. Providing consistent, reproducible and up-to-date information on various forest parameters is the main advantage of using the latest remote sensing data and methods. New possibilities are given by the airborne digital sensor ADS80, which records the entire country during the vegetation season every six years. This paper presents a robust methodology of classifying tree species in different study areas. The obtained accuracies for beech, ash, Norway spruce, Scots pine, larch, willow and silver fir are in average 71–85%, but lower for other deciduous tree species. These are mainly less dominant tree species within a study area such as maple and birch. A small sample data set and shadows of other neighboring trees seem to be the main reasons for this. Based on the experiences made in this study, a countrywide classification of tree species has become more feasible. The usage of airborne digital sensor ADS80 data in combination with a high degree of automation from the developed methods will enable the generation of countrywide products on the distinction of coniferous and deciduous tree species until 2015.