To investigate responses to longitudinal torsion and the effect of temperature on the torsional stiffness of superelastic nickel titanium archwires, eight batches of rectangular wires were tested at 18, 27, 37, and 40 degrees C ambient temperature. The wires were twisted 25 degrees and studied in deactivation. The resulting torque-twist diagrams show that only half the wires had discernible deactivation plateaus at body temperature. The plateaus were generally narrow (1 to 3 degrees) and started at 20 to 23 degrees of torsional twist. The clinical significance of these deactivation plateaus is debatable. Only one wire had a well-defined plateau that was fairly wide (6 degrees) and started at a lower level of twist (17 degrees). The wires without plateaus when deactivated from 25 degrees of twist were retested at body temperature. All exhibited deactivation plateaus subsequent to activating twists of 45 and 60 degrees, and the plateaus became more distinct as the degree of prior activation increased. This indicates that the stress imparted on the alloys by 25 degrees of activating twist is insufficient to induce martensitic transformation at body temperature. As prescriptions advocate bracket pretorque of less than 25 degrees for a maxillary central incisor, the clinical relevance of alloys requiring large activations before they demonstrate deactivation plateaus is questionable. Half the wires tested were markedly influenced by ambient temperature changes; the other half were relatively insensitive to temperature. Responsiveness to thermal stimuli seemed closely related to superelastic tendency.