The heteronuclear 15N-NOE experiment was extensively tested with respect to statistical and systematic experimental error. The dependence of signal intensity in the NOE experiment and in the reference experiment on the saturation and relaxation time was experimentally investigated. The statistics of the experimental values were accessed by numerous repetitions of identical set-ups. As a model system a protein of typical size for NMR studies was chosen, i.e., a 120 amino acid residues containing fragment of the F-actin binding gelation factor (ABP-120). The fragment exhibits fast dynamics that are accessible with the 15N-NOE experiment with various amplitudes. The results of this study show that commonly used parameters are only adequate for accurate measurement of motions with moderate amplitude. Highly flexible parts require longer delay times and thus more experimental time than commonly used. On the other hand, a qualitative or semi-quantitative assessment of a protein's mobility on fast times scales can be obtained from rapidly recorded experiments with unusual short delay times. The findings of this study are of equal importance for highly accurate measurement of the 15N-NOE as well as for quick identification of mobile or even unstructured residues/parts of a protein.