Terrestrial laser scanner record not only the 3D coordinates per point, but also the strength of the receiving signal. In addition to the target scattering characteristics, the signal is strongly influenced by the measurement geometry, the instrumental and the atmospheric effects. By detailed calibration of these additional influences, the target scattering characteristics can be derived from the receiving signal. Above all, the instrumental effects vary strongly for different TLS models, so that a direct comparison of its receiving signals is not possible. By measuring objects of known reflectivity, calibration curves can be created for each scanner. If these calibration curves are applied to intensities of different TLS models, the generated reflectivities can be compared with each other. In this thesis, a dataset of a forest is used to show how the structures of a tree (trunks, branches, needles and leafes) can be distinguished by simply comparing reflectivities of different wavelengths. As a result, a classification of the two classes trunks and needels/leaves by their values of a modified NDVI provides a overall accuracy of 74%. By only using single echoes a overall accuracy of almost 90% is achieved. This is based on the fact that single echoes have more distinct reflectivities and therefore the NDVI of the individual structures differs more clearly. A classification based on the reflectivities at a wavelength of 1.5 m provides a overall accuracy of 90% and for single echoes even 94%. Whereas a distinction of tree structures based on the reflectivity at 1.0 m is not possible. This demonstrates that although it is possible to compare and combine reflectivities of different TLS and wavelengths, it is not necessary for the purpose of classification.