The bend and free recovery (BFR) test according to ASTM F2082 is a standard method to determine the transition temperatures of Nitinol shape memory alloys (SMAs). Unfortunately, this standard method is limited to SMA wires which are straight in its trained shape. Thus, the standard BFR test is not suitable for thermomechanical characterization of curved Nitinol SMA wires which should serve as actuators in cochlear implants in future. We developed a modified BFR measurement setup to determine the active austenite finish (AF) temperature of these very thin wires (Ø100 μm). The active AF temperature specifies the completion of the shape recovery upon heating. A parametric study of the measurement setup was carried out to investigate the influence of the heating rate on the observed active AF temperature and to verify the repeatability of the measurement setup. First, the curved wire was straightened in a cold water bath before inserting it into a water bath that is gradually heated from 5 °C to 45 °C. The shape change of the previously straightened wire was then recorded throughout the experiment using a digital microscope. Five different heating rates were employed: 0.25 K/min, 0.33 K/min, 0.5 K/min, 1 K/min as well as an unregulated maximum heating rate achievable of approximately 1.5 K/min. Furthermore, an investigation on the test-retest reliability was performed with three wires by repeating the experiment ten times with each wire. The results of this study revealed no influence of the heating rate on the thermomechanical response of the wires. Based on data from this study, a regulated heating rate of 1 K/min is suggested for future investigations, as this reduces the duration of the measurement from four hours to less than an hour. The values obtained from each wire through the test-retest reliability investigation showed a standard deviation of 1.9 K, 1.1 K and 2.1 K respectively. Our developed measurement setup demonstrates appropriate repeatability of the measurements.