Khorasan province bieng located in an arid and semiarid part of Iran, has often experienced drought during the recent years. Occurrence of consequent droughts during the last few years, has shown that drought prediction is a subject that deserve more attention. One way to achive such goal is modeling the rainfall. In this research, annual rainfall data of the eleven synoptic stations of Khorasan province from 1970 to 2002 have been used. Seasonal autoregressive integrated moving average method (SARMA) was used for modeling the seasonal rainfals of these stations. Based on this model, the amount of rainfall for spring, fall and winter was predicted. Anomally zoning was also prepared for Khorasan province.