The choice of the most suitable orthodontic wire for each stage of treatment requires estimation of the forces generated. In theory, the selection of wire sequences should initially utilize a lower flexural rigidity; thus clinicians use smaller round cross-sectional dimension wires to generate lighter forces during the preliminary alignment stage. This assessment is true for conventional alloys, but not necessarily for superelastic nickel titanium (NiTi). In this case, the flexural rigidity dependence on cross-sectional dimension differs from the linear elasticity prediction because of the martensitic transformation process. It decreases with increasing deflection and this phenomenon is accentuated in the unloading process. This behaviour should lead us to consider differently the biomechanical approach to orthodontic treatment. The present study compared bending in 10 archwires made from NiTi orthodontics alloy of two cross-sectional dimensions. The results were based on microstructural and mechanical investigations. With conventional alloys, the flexural rigidity was constant for each wire and increased largely with the cross-sectional dimension for the same strain. With NiTi alloys, the flexural rigidity is not constant and the influence of size was not as important as it should be. This result can be explained by the non-constant elastic modulus during the martensite transformation process. Thus, in some cases, treatment can begin with full-size (rectangular) wires that nearly fill the bracket slot with a force application deemed to be physiologically desirable for tooth movement and compatible with patient comfort.