Temperature is one of the most important climatic variables being effective in specification and distribution of other variables of climate and is one of the main components of the zoning and classification. One way to planning and management is based on the knowledge of the thermal properties and their coverage area. In this study, using cluster and discriminant analysis multivariate methods, temperature zoning in the North West of the country was done. Using Kriging, 720 monthly minimum and maximum temperature maps with 8×8 km dimensions were created. Finally by removing cells out of range, 2436 cells were obtained, and a matrix database was created in the R form in 2436 rows (cells) multiplied 24 columns (variables). By applying cluster analysis four different temperatures in the North West area using Euclidean distance and Ward methods was identified. Discriminant analysis was used for classification and testing the accuracy of the cluster analysis. Based on the results of discriminant and cluster analyses, the groups differ only in %1.6 of the area. This result of the cluster and discriminant analyses also show that there is not a very commanding difference. This can be confirmed with the results of clustering. Each thermal zone was named on the basis of the comparison of each region with Iran’s average temperature (18 T^◦C). These temperature areas are: Very cold area, cold area, semi-cold with cooler day and warmer night and semi-cold with warmer day and cooler night. The identified geographic areas temperature patterns which are more consistent with altitudinal levels unraveled that cluster analysis can be considered as a useful tool for zoning thermal districts.