The choice of reference evapotranspiration (ETo) estimation methods and general circulation model (GCM) are crucial for projecting water deficit under a changing climate. Standardized Precipitation Evapotranspiration Index (SPEI) derived from water deficit also varies with the choice of GCM and ETo estimation methods. In this study a variance-based global sensitivity analysis was used to estimate relative sensitivity of projected changes in future water deficit (P-ETo) and SPEI to the choice of GCM and ETo estimation methods over parts of the Indian subcontinent. For evaluating the change in water deficit and droughts, 7 GCMs and 11 ETo methods were analyzed for two distinct periods i.e. 2030-2060 and 2070-2100 compared to the baseline (1951-1980). The 11 ETo methods were grouped into 4 major categories namely based on temperature, radiation, mass transfer and combination methods. Moreover, based on the ETo categories, a non-parametric Mann-Whitney was performed to quantify robust changes under a warming climate. Results show that changes in future water deficit and droughts varies with regions and seasons. Overall, changes in water deficit droughts are more inclined to the choice of ETo method, while the GCM-ETo interaction effects are more prominent in some regions. Results also showed that within an individual ETo category, individual ETo methods do not necessarily agree on the magnitude/direction of change in projecting water deficit and SPEI for future conditions. This has important implications towards selection of appropriate ETo estimation for drought analysis in data scarce regions under a changing climate. Results of this study indicate, the role of proper ensemble formation of GCMs and ETo estimation methods based on seasons and regions, to develop a robust range of future conditions for water resources planning.