Assessing climate change impact on hydrology and biomass yield of a tropical river basin is necessary for formulating a suitable coping strategy. A hydrological modelling framework was established for Subarnarekha river basin through calibration and validation of SWAT model with the observed river discharge at four gauging locations. The global sensitivity analysis was performed to establish the robustness and stability of the model to reproduce outputs resultant to various extremes of precipitation and other weather events. The 5-GCMs, namely HadGEM2-ES, MIROC-ESM, NCAR-CCSM4, CSIRO-MK3.6, and CESM1-CAM5 were analyzed and HadGEM2-ES was selected based on its performance with respect to ensemble of remaining 4 GCMs (p ≤ 0.05). The AR5-HadGEM2-ES monthly database (1 km2 resolutions) was used to generate bias factor for preparing daily climatic variables in the future projection which was employed as a forcing to simulate river discharge and other hydrological fluxes for the future climate scenario. Increase in precipitation from 14 to 36%, the runoff increased from 39.7 to 104.1%. When compared to the present period, an increase in monsoon runoff is expected 100 and 200% during 2030 and 2080, respectively. Higher degree of uncertainty in expected runoff volumes was expected in mid- and far future time period. Intra-RCP scenarios analyses indicated a significant surge in monsoon season runoff volume during 2070 and 2080 and insignificant surge during 2030 and 2050. As the basin constitute a rice ecosystem, the simulation approach involved a detailed accounting of the operational activities of a rice to realistically capture its manifestation in the annual accrual of actual evapotranspiration in the simulated hydrological fluxes. During monsoon season, potential evapotranspiration (PET) and actual evapotranspiration (AET) increased by 2.2–12.7% and 1.0–9.0%, respectively. Following simulation output, it was found that the basin was expected to suffer a crop biomass loss of 2 to 3% due to climate change. Simple yield response technique to estimate biomass yield losses was found to be promising alternatives to complex crop modeling studies.