Abstract: This study presents a model that can be used in planning a release policy for Kanji hydropower station. This will bring about an optimal release policy. The monthly Reservoir Storage for Kanji hydropower system was plotted to determine the degree of seasonality and to identify outliers. The sample auto-correlation when plotted showed the series to be non-stationary. The Box-Jenkins univariate procedure was adopted for model development. The method of differencing was employed to make the series stationary. The auto-correlation and the partial auto-correlation parameters of the stationary series were used to identify the appropriate model. The decaying pattern of the autocorrelation and partial autocorrelation indicated the need for a mixed model. The mixed model developed was used to forecast storage values. The forecast values were found to compare favourably with the actual data. The model is a short-term forecast method that can be relied for the daily operations of the hydropower system.