Predicting the amount of available water is one of the important and effective factors in programming and managing the water resources and its estimation during various temporal scales has great importance in agricultural planning. Time series is one of the studying methods for predicting. Various models and approaches have been proposed to achieve this purpose; including ARIMA and AR,MA temporal series models. In this research, the performance of each mentioned models during the estimation of future amounts Zayandehrood River since 2006 to 2012 was studied as 72 monthly data series. The results showed that in comparison to all models, AR12 has a more desirable performance and simulates the time series changing trend with less error.