The objective of this paper is to find an appropriate Seasonal Auto-Regressive Integrated Moving Average (SARIMA) model for fitting the monthly discharge of a karstic spring located at the North of the city of Serres (Agios Ioannis, Mount Menikio) by considering the minimum of Akaike Information Criterion (AIC). Box- Jenkins methodology applies models to find the best fit of a timeseries to past values of this timeseries, in order to make forecasting and consists of a four-step iterative procedure: identification, estimation, diagnostic check and forecasting. Timeseries analysis and forecasting of hydrological parameters such as spring discharge may be useful in decision making and optimum water resources usage. In this study, monthly discharge measurements are analysed. Initial data are firstly transformed to normal and stationary using differencing methods. Autocorrelation and Partial Autocorrelation functions are calculated to determine the order of Autoregressive and Moving Average parameters and residuals are then checked to show the “white noise”. The spring discharge data are forecasted based on the selected model up to 2008 and are then compared with measured values. The timeseries model SARIMA (2,1,1)(1,0,1)12 could be used in monthly discharge forecasting at a short time (upcoming one year) with a simple and explicit model structure in order to help decision m akers to establish priorities in terms of water demand management. Finally, the corr elation coefficient between the observed and fitted data is essentially high, while the absolute and relative errors are significantly low.