The glass transition temperature (Tg) is an important property that influences the processing and textural characteristics of candy. Measurement of Tg is done by comparably expensive and complex instruments. In this study, we tested a new system which uses capacitance thermal analysis (CTA) of candy trapped between stainless plates, as the system is caused to heat at an uncontrolled rate. The data of capacitance as a function of temperature were processed by a genetic algorithm (GA), and fitted to a three-section model to determine Tg. Tg of the candies were independently measured by DSC as a reference. The results showed that when the Tg of the candy was below ∼15 °C, the measurement from the GA-CTA was higher (2–3 °C) than that from DSC. However, if Tg of the candy was higher than 15 °C, the two methods gave similar values. GA based CTA provides a feasible new way to measure phase transitions in candies with relatively inexpensive equipment, and with less need for user interpretation of data.