In recent years, climate change is an effective factor that has important effects on the drought extension, changes in cropping patterns and negative environmental impacts. In this regard, estimation of maximum air temperature has more importance in climatological, agricultural studies and water resources management, therefore the selection of a comprehensive approach is necessary for air temperature estimation. One of the options is to use statistical (regression) methods, which two factors affect the efficiency of method: effective variables of method and type of statistical method. In this research, effective variables of method were investigated from climatological aspect and study of second factor was based on comparison of classical and fuzzy regression methods using data of some meteorological station of Iran. In different climates, the effects of meteorological variables on the maximum air temperature are not the same and based on coefficient of determination, meteorological variables of arid and semiarid regions have more similarities. Mean air temperature has the maximum coefficient of determination in all climates. In case of the method type, Banar Anzali – Ramsar and the other stations have better performance using non-symmetric and symmetric fuzzy regression, respectively. Based on criteria such as RMSE, fuzzy regression in comparison to the classical regression has more efficiency.