Protein structures are often represented as seen in crystals: (i) rigid macromolecules (ii) with 3-states representations, namely helices, sheets and coils. Nonetheless, both information are incomplete because (i) proteins are highly dynamical macromolecules and (ii) the description of protein structures can be more precise. Regarding these two points, we have analysed and quantified the stability of helices by considering α-helices as well as the other two other helical local conformations, namely the 310-helices and the rare π-helices. Molecular dynamics simulations of 150 ns were performed on a large set of 169 representative protein domains with three independent simulations for each protein chain. The local protein conformations were studied during each simulation and analysed. Classical flexibility index (B-factor) was confronted to MD flexibility index (RMSf) and protein blocks (PB) assignation to characterize distorsions. Helical regions were classified according to their level of helicity from highly to none. For the first time, a precise quantification showed the percentage of rigid and flexible helices underlying unexpected behaviours. For instance, only 76.4% of the residues associated to α-helices keep this conformation, while this tendency drops to 40.5% for 310-helices and is never seen for π-helices. This analysis highlight the carefulness of considering protein structures as dynamic entity rather than static ones.