Background Many studies have shown some relationships between thermal perception (including thermal sensation and thermal comfort) and human physiological parameters, such as brain signals. However, further research is still needed on how these parameters can help recognize the state of a human’s personal thermal perception. Objective This study aims to investigate the potential of using fNIRS brain signals to evaluate and predict personal thermal perception and cognitive performance in a steady-state temperature. Methods The present study investigated changes in the fNIRS signal during ambient temperature manipulation. Thirty healthy young individuals were selected as the subjects, and they were exposed to two steady temperatures of 28.8 and 19 °C. After acclimatizing to either temperature, the oxy/deoxy-hemoglobin changes of the prefrontal cortex (PFC) were measured in both rest and cognitive task states using 16-channel fNIRS. Results Results showed that exposure to different temperatures was significantly associated with the brain signals recorded during the task state. Many significant correlations were discovered between fNIRS signals and thermal perception indices. Furthermore, subjects’ performance changes led to changes in the fNIRS signals. Logistic regression showed that fNIRS can determine whether a person is thermally comfortable or uncomfortable.