The Predicted Mean Vote (PMV) created by Fanger in the 70's is the most common model for estimating the thermal sensation in a group of individuals. The PMV contains some discrepancies related to the thermal reality of the environments where it is applied; thus, several researchers have developed alternative models over the past years in order to reduce these differences. This research aimed at analyzing thermal comfort conditions in Brazil by applying the PMV model and alternative models through a statistical discriminant analysis. By using the ASHRAE Global Thermal Comfort Database II, the environmental and personal variables of thermal comfort were applied in order to calculate the responses of each model and thus compare them to the responses of thermal sensation reported by people in four Brazilian cities: Brasília, Recife, Maceió and Florianópolis. The models used in this research were selected by applying the Preferred Reporting Items for Systematic Review and Meta-Analysis (PRISMA) method. A discriminant analysis was performed in order to classify people according to their thermal sensation in different groups, through the development of discriminant functions; this procedure allowed a better visualization of these groups. With the results, the discrimination of individuals was verified according to their thermal sensations, where the models were right to classify individuals into groups with 96.1% for Brasília and Recife; 99.1% for Florianópolis and 99.8% for Maceió, thus showing the effectiveness of the models.