Colour is an important criterion for wheat flour quality. A strong pigmentation of the endosperm by carotenoids is desirable in durum wheat (Triticum durum) since a bright yellow colour is a quality factor of pasta. Contrary, flour of bread wheat (T. aestivum) should be as white as possible. Hence, a minimum pigmentation is sought after for baking products. The determination of yellow endosperm pigmentation by standard methods requires both chemicals, respective lab equipment and time. The aim of this thesis was (i) to develop a near infrared spectroscopy (NIRS) calibration for wheat flour for the prediction of yellow pigments and (ii) to check if it is possible to differentiate among different wheat species (T. aestivum, T. durum, T. monococcum, T. carthlicum, T. karamyschevii, T. turanicum) on basis of their spectral data. In the present study Fourier transform NIRS was used to develop a prediction model for yellow endosperm pigmentation across various wheat species to allow a faster, cheaper and chemical free selection of genotypes. The obtained prediction model is robust enough for selection, however, misses analytical accuracy (R=0.82-0.88). Samples spiked with various concentrations and types of carotenoids were used to identify infrared wavelength regions which reacted significantly to the respective carotenoids. Principal component analysis was applied to the spectra for qualitative analysis and revealed a significant clustering of genotypes according to wheat species. The grouping of genotypes, respectively wheat species, was, however, mainly influenced by kernel hardness and particle size and not by the yellow pigmentation.