Abstract FTIR spectroscopy was used to distinguish between beech (Fagus sylvatica L.) trees grown at five different sites; one in middle Germany close to Göttingen (forest district Reinhausen), three located in the southwest (two in Rhineland-Palatinate: forest districts Saarburg and Hochwald, and one in Luxembourg), and one in North-Rhine Westfalia. Detailed investigation of the spectra in the fingerprint region (1800–600 cm-1) revealed 16 distinct peaks and shoulders, most of which were assignable to wavenumbers previously shown to represent wood compounds. Differences in peak heights and peak ratios indicated differences in wood composition of beech trees from different sites. To determine if the wood of individual trees could be distinguished, principal component analysis (PCA) and cluster analysis were performed using FTIR spectra as input data. With both PCA and cluster analysis, trees from four of the five different sites were separated. It was not possible to distinguish between trees from Saarburg and Hochwald, where similar edaphic and climatic conditions exist, while wood spectra from trees from all other areas clearly segregated. Wood collected at different positions in the stem (bottom, crown, center and outer year rings) of trees grown at the same site was not distinguishable. Therefore, FTIR spectral analysis in combination with multivariate statistical methods can be used to distinguish wood of trees from different growth habitats. Extension of this method to other species may be of great interest for wood certification, as it may be possible to distinguish wood, of a given species, originating from different regions.