The spatial and temporal variability of air temperature, precipitation, actual evapotranspiration (AET) and their related water balance components, as well as their responses to anthropogenic climate change, provide fundamental information for an effective management of water resources and for a proactive involvement of users and stakeholders, in order to develop and apply adaptation and mitigation strategies at the local level. In this study, using an interdisciplinary research approach tailored to water management needs, we evaluate the past, present and future quantity of water potentially available for drinking supply in the water catchments feeding the about 2.3 million inhabitants of the Turin metropolitan area (the former Province of Turin, north-western Italy), considering climatologies at the quarterly and yearly timescales. Observed daily maximum surface air temperature and precipitation data from 1959 to 2017 were analysed to assess historical trends, their significance and the possible cross-correlations between the water balance components. Regional climate model (RCM) simulations from a small ensemble were analysed to provide mid-century projections of the difference between precipitation and AET for the area of interest in the future CMIP5 scenarios RCP4.5 (stabilization) and RCP8.5 (business as usual). Temporal and spatial variations in recharge were approximated with variations of drainage. The impact of irrigation, and of snowpack variability, on the latter was also assessed. The other terms of water balance were disregarded because they are affected by higher uncertainty. The analysis over the historical period indicated that the driest area of the study region displayed significant negative annual (and spring) trends of both precipitation and drainage. Results from field experiments were used to model irrigation, and we found that relatively wetter watersheds in the northern and in the southern parts behave differently, with a significant increase of AET due to irrigation. The analysis of future projections suggested almost stationary conditions for annual data. Regarding quarterly data, a slight decrease in summer drainage was found in three out of five models in both emission scenarios. The RCM ensemble exhibits a large spread in the representation of the future drainage trends. The large interannual variability of precipitation was also quantified and identified as a relevant risk factor for water management, expected to play a major role also in future decades.