The transformation between martensite and austenite is characterized by four transformation temperatures: martensite start temperature (Ms), martensite finish temperature (Mf), austenite start temperature (As) and austenite finish temperature (Af). In actuator design that relies on the shape memory effect, it is important to obtain an accurate measure of these transformation temperatures, especially As and Af. Several methods of determining these temperatures have been reported, but their accuracy and coherence are not clear. Three methods were used to measure the transition temperatures of NiTi wire under different heat treatment conditions: differential scanning calorimetry (DSC); an electrical resistance method, which uses a sudden change in resistance as an indication of transformation; and an applied loading method, where a macroscopic change in displacement indicates the transformation. The results show that the transition temperatures measured by DSC do not correspond to those measured by the other two methods, which are similar. The applied loading method is the most effective for providing practical information about the stress-dependent transformation temperatures. The electrical resistance test gives clearly determined points for Ms and Mf in the cooling resistance-temperature curve, but As and Af are not clearly identifiable in heating process.