**Aims** The 20th century has been marked by dramatic changes in land use, disturbance regimes and climate, which have interacted to affect global ecological patterns and dynamics, including changes in the extent, composition and structure of forest cover. Although much research has highlighted dramatic, short-term ecological change, on-going trends of land-use change and climate change began more than a century ago. Consequently, quantifying and understanding long-term (e.g. centennial) ecological change is critical to contextualizing recent patterns and processes. Here we document changes in the extent, position and composition of sub-alpine forests over the past century in eastern Switzerland. **Location** Davos region of the Swiss Alps, eastern Switzerland. **Methods** Position of tree line, forest cover and forest composition were evaluated using a unique combination of Object-Based Image Classification of an historical (1909) map, recent (2009) aerial photography and repeat terrestrial photography to minimize the inherent bias of each data source, while providing the most robust representation of long-term ecological change. **Results** Over the past century total forest cover expanded by 64.6% and the position of sub-alpine tree line increased on all aspects. Total forest cover also increased at the highest and lowest elevations on all aspects. Dominance of European larch increased at the highest elevations, but decreased at the lowest elevations, where it was replaced by Norway spruce. These patterns suggest land use has been the most important driver of forest change over the past century. **Conclusions** Major changes in the extent, structure and dynamics of sub-alpine forests in the Alps initiated earlier than previously documented and most change occurred prior to the middle of the 20th century. Furthermore, these changes were likely driven primarily by changes in land use, rather than by changes in climate. A combination of data sources and methodological approaches, such as those of the current study, provides a clearer view of long-term changes and minimize the biases associated with any single data source or methodology.