Climate change will alter the inflows into the dam in the future; thus, the balance between water supply and water availability will directly impact the water levels and indirectly affects dam safety. Therefore, estimating the future inflows and reservoir water content can help the operators. In this study, a machine learning Wavelet-MultiLayer Perceptron (W-MLP) method is applied to estimate monthly future projections (2016-2100) of the inflows into the reservoir. The methodology is tested for one of the main water supply reservoirs in Ankara, which distributes annual 142 hm 3 water. The EURO-CORDEX database is used to obtain Regional Climate Model (RCM) simulations of RCA4 (12.5 km) from two different Global Circulation Models (GCMs), MPI-ESM-LR and IPSL-CM5A-MR, under two Representative Concentration Pathways (RCPs) (RCP 4.5 and RCP 8.5) scenarios. The monthly W-MLP models are independently trained and tested for each data set (observed data and GCM outputs). The GCM scenario results indicate a shift in monthly hydrographs for both RCPs projections with a reduction in inflows which will directly change the operation of the reservoir. The daily HEC-ResSim model mimics future water content and releases. According to the results, the annual reduction expected in the future inflows scenarios varies between -3 % to -13% under the RCPs, and the effects on annual reservoir water content are much higher (between -21 % and -37 %). These findings can be used in different risk assessment metrics (reliability, resilience, and vulnerability) to estimate the future effects of dam safety.