One of the research subject in climate is study and identification of frost and its temporal and spatial distribution in each geographic region which threaten agricultural products in different parts of the world. Maragheh is located at a mountainous area of the country and it has high altitude. So, frost damages crops and horticultural plants heavily each year. The economic reliance of the people of this region upon agriculture requires the careful study of features of this phenomenon. In this research, five agro climatic indices, i.e. first fall frost date, last spring frost date, growing season (frost free season), early fall frost date, late spring frost date were considered. First, using minimum temperature data in daily scale (1984-2015), frost indices were extracted. Then, using Matlab software, among 17 probability distribution, best distribution for each frost index was determined. Results showed that Logistic, Inverse Gaussian, Extreme Value, Log Logistic distribution were identified as the best fitted distributions for first frost, last frost, early fall frost and late spring frost, respectively. Weibull was selected as appropriate distribution for growing season. Afterwards, using selected distributions in different probability levels, probable dates of the studied indices in different probability levels and return period was calculated in Matlab. It was found that every 5 year, the date of first frost and last frost will be on 7 November and 28 March. Early fall frost and late spring frost will occur on 5 November and 17 April. The length of growing season will be 201 days.