Forcing watershed models with downscaled climate data to quantify future water regime changes can improve confidence in watershed planning. The Soil Water Assessment Tool (SWAT) was calibrated (R2 = 0.77, NSE = 0.76, and PBIAS = 7.1) and validated (R2 = 0.8, NSE = 0.78, and PBIAS = 8.8) using observed monthly streamflow in a representative mountainous watershed in the northeastern United States. Four downscaled global climate models (GCMs) under two Representative Concentration Pathways (RCP 4.5, RCP 8.5) were forced. Future periods were separated into three 20-year intervals: 2030s (2031–2050), 2050s (2051–2070), and 2070s (2071–2099), and compared to baseline conditions (1980–1999). Ensemble means of the four GCMs showed an increasing trend for precipitation with the highest average increase of 6.78% in 2070s under RCP 8.5. Evapotranspiration (ET) had increasing trends over the 21st century with the 2030s showing greater increases under both RCPs. Both streamflow (4.58–10.43%) and water yield (1.2–7.58%) showed increasing trends in the 2050s and 2070s under both RCPs. Seasonal increases in precipitation were predicted for most months of spring and summer. ET was predicted to increase from Spring to early Fall. Study results demonstrate the potential sensitivity of mountainous watersheds to future climate changes and the need for ongoing predictive modeling studies to advance forward looking mitigation decisions.