Global climate change and urban heat island intensification are making the outdoor environment more thermally uncomfortable during hot weather, and more dangerous during heat waves. This has generated considerable interest in the development of models to estimate the effect of heat on human bodies. However, it is not clear that which model provides the most accurate and precise results and which method could appropriately analyze the data. This study, conducted in Shanghai, China, compared the outputs of three of the most accepted human energy budget models with a database of 509 thermal sensation votes under various thermal conditions. We analyzed the results in three ways, including linear regression, curvilinear regression, and multinomial logistic regression assessment. Results showed that because thermal sensation votes are ordinal data, so the first two methods have theoretical limitations, while the latter method was determined to be the most appropriate for the data type of the three models tested. Further, in hot and humid climate area, we found COMfort FormulA (COMFA) is more accurate than Physiologically Equivalent Temperature (PET) and Universal Thermal Comfort Index (UTCI), thus it has more potential for use by urban designers in improving outdoor thermal comfort.