The present study deals with the relevance of using mobility-averaged dipolar couplings for the structure refinement of flexible proteins. The 68-residue protein p8MTCP1 has been chosen as model for this study. Its solution state consists mainly of three alpha-helices. The two N-terminal helices are strapped in a well-determined alpha-hairpin, whereas, due to an intrinsic mobility, the position of the third helix is less well defined in the NMR structure. To further characterize the degrees of freedom of this helix, we have measured the dipolar coupling constants in the backbone of p8MTCP1 in a bicellar medium. We show here that including D(dip)HN dipolar couplings in the structure calculation protocol improves the structure of the alpha-hairpin but not the positioning of the third helix. This is due to the motional averaging of the dipolar couplings measured in the last helix. Performing two calculations with different force constants for the dipolar restraints highlights the inconstancy of these mobility-averaged dipolar couplings. Alternatively, prior to any structure calculations, comparing the values of the dipolar couplings measured in helix III to values back-calculated from an ideal helix demonstrates that they are atypical for a helix. This can be partly attributed to mobility effects since the inclusion of the 15N relaxation derived order parameter allows for a better fit.