Groundwater extraction is used to alleviate drought in many habitats. However, widespread drought decreases spring discharge and there is a need to integrate climate change research into resource management and action. Accurate estimates of groundwater discharge may be valuable in improving decision support systems of hydrogeological resource exploitation. The present study performs a forecast for groundwater discharge in Aquifer?s Cervialto Mountains(southern Italy). A time series starting in 1883 was the basis for longterm predictions. An Ensemble Discharge Prediction(EDis P) was applied, and the progress of the discharge ensemble forecast was inferred with the aid of an Exponential Smoothing(ES) model initialized at different annual times. EDisP-ES hindcast model experiments were tested, and discharge plume-patterns forecast was assessed with horizon placed in the year 2044. A 46-year cycle pattern was identified by comparing simulations and observations, which is essential for the forecasting purpose. ED is P-ES performed an ensemble mean path for the coming decades that indicates a discharge regime within ± 1 standard deviation around the mean value of 4.1 m~3 s~(-1). These fluctuations are comparable with those observed in the period 1961-1980 and further back, with changepoints detectable around the years 2025 and 2035. Temporary drought conditions are expected after the year 2030. Groundwater extraction is used to alleviate drought in many habitats. However, widespread drought decreases spring discharge and there is a need to integrate climate change research into resource management and action. Accurate estimates of groundwater discharge may be more in improving decision support systems of hydrogeological resource exploitation. The present study performs a forecast for groundwater discharge in Aquifer? s Cervialto Mountains (southern Italy). A time series starting in 1883 was the basis for longterm predictions. An Ensemble Discharge Prediction (EDis P) was applied, and the progress of the discharge ensemble forecast was inferred with the aid of an Exponential Smoothing (ES) model initialized at different annual times. EDisP-ES hindcast model experiments were tested, and discharge plume-patterns forecast was assessed with horizon placed in the year 2044. A 46 -year cycle pattern was identified by comparing simulations and observations, which is essential for the forecasting purpose. ED is P-ES performed an ensemble mean path for the coming decades that indicates a discharge regime within ± 1 standard deviation around the mean value of 4.1 m ~ 3 s -1. These fluctuations are comparable with those observed in the period 1961-1980 and further back, with changepoints detectable around the years 2025 and 2035. Temporary drought conditions are expected after the year 2030.