A new method of characterizing the curvature change in thermally adaptive fibers is introduced in this paper. Based on the same principle as bi-metallic strips commonly found in thermostats, multi-component polymer fibers can be created to change their geometrical form in response to a temperature change. This works by creating fibers from two or more materials that have a mismatched Coefficient of Thermal Expansion (CTE). A temperature change leads to a change in curvature of these fibers. When fibers interact in an insulation batting structure, a temperature change leads to a thickness change in the insulation. While these fibers have visually been observed to function, there was no method to quantitatively characterize their curvature performance. This paper introduces a method that can be used to quantify fiber performance by tracking change in curvature over a specific temperature range. This is accomplished by suspending fibers on the surface of a liquid bath and changing the bath temperature. Digital images of the fiber are taken at different temperatures and analyzed using software to determine the radius of curvature. Absolute change in curvature was found to be as high as 0.5% per degree °C from 20 to −20°C for certain samples. A trend was also noted between higher initial curvature and lower overall performance. Digital image correlation was further used to investigate the time-dependence relationship of fiber curvature. Future experiments can be performed with this setup to characterize and compare curvature change performance of different fibers accurately.