Rapidly depleting groundwater in western Indo-Gangetic Plains (IGP) is a major threat to food security in South Asia. Conventional tillage-based and flood irrigated puddled transplanted rice (PTR) is a major contributor to faster depleting aquifers. Urgent actions are therefore warranted to develop alternate productive, profitable, water and N-use efficient rice production practices for rice-wheat (RW) cropping system. Conservation agriculture (CA) based direct-seeded rice (DSR) has been advocated as a potential alternative to PTR. Further, bundling CA with precision water and N management using sub-surface drip irrigation (SSD) has demonstrated significant benefits over CA-based flood irrigation (FI). However, for more efficient use of water, water budgeting is needed which is a challenging task as it requires expensive tools, and time, and efforts. Information about complete water budgeting in high water demanding crops like rice grown under CA-based SSD, FI, and PTR are not available. We deployed HYDRUS-2D model for estimating water budgeting of rice under CA+ (CA-based SSD), CA-based FI, and PTR-based systems. The objective of our study was to calibrate and validate the HYDRUS-2D model to simulate water dynamics in rice grown under CA-based SSD and FI compared to PTR and to design water and N- use efficient production practices for rice cultivation in western IGP. Five treatments comprised of PTR+FI with 120 kg N ha−1 (PTR), zero-till direct-seeded rice (ZTDSR)+FI without N (ZT-N0), ZTDSR+FI with 100% of N recommended dose (ZT-N100), ZTDSR+SSD without N (SSD-N0), and ZTDSR+SSD with 100% of N-recommended dose (SSD-N100) were compared. The result showed that the HYDRUS-2D model satisfactorily simulated the soil moisture content with low root mean square error (RMSE) (0.014–0.028), high coefficient of determination (74–92%), and model efficiency (59–87%) during the simulation period (80 days: 35–114 days after sowing). The highest grain yield (7.18 t ha−1) was observed in the PTR treatment, which was statistically similar to SSD-N100 (6.54 t ha−1) and significantly higher than ZT-N100. During the simulation period, PTR plots received 131.7 cm of water (rainfall + irrigation) which was 27.3% and 50.1% higher than ZT-N100 and SSD-N100 plots, respectively. Out of the cumulative water applied, PTR transpired only 18.4% of applied water, compared to 24% in ZT-N100 and 36.3% in SSD-N100. Interestingly, SSD-N100 plots recorded 20.6% and 23.5% less evaporative loss and 45.0% and 66.0% less water loss by deep drainage than ZT-N100 and PTR, respectively. Thus, conversion to CA+ system with 100% N-recommended dose saved 50.1% and 31.3% of water, and consequently attained 2.0 and 1.45-times higher biomass water use efficiency than PTR and ZT-N100, respectively. Based on the results, CA-based SSD could be recommended for precise utilization of water and to curtails the unproductive water loss components such as evaporation and deep drainage.