This study examined a representative college activity centre in China’s cold regions and conducted a field measurement study and questionnaire analysis on the indoor thermal environment in winter. Data fitting and comparative analysis were performed on the predicted mean vote (PMV) calculated from measured environmental parameters and the seven-point vote value actual mean vote (AMV) based on the questionnaire. According to the least square method, the corresponding human adaptive coefficient λm in the motion state and λs in the stationary state were obtained. Based on the research, the thermal comfort survey results were reported for the two states for recreational spaces of colleges and universities. These results suggest that the AMV and PMV do not coincide in the two states. The actual thermal sensation is less sensitive than predicted through the sensitivity analysis for the two states. The respondents’ actual thermal sensation is less sensitive when exercising than stationary state. Moreover, both PMV and AMV in motion state are higher than those in stationary state, illustrating that the state of motion has a greater influence on thermal sensation. This research provides a theoretical basis and data reference for the low energy design of recreational spaces of colleges and universities.