Dry direct seeding is an alternative rice production method that reduces 50-60% irrigation water required compared to the conventional method. The present study was undertaken with a view to optimize irrigation requirement for dry direct seeded rice. The experiment involved seven water management levels and two rice varieties in a split-plot design allocating water management levels in main plot and rice variety in subplot. The treatments were replicated thrice. The water management levels are (i) puddled transplanted rice with full irrigation (PTR-CI), (ii) puddled transplanted rice with alternate wetting and drying (AWD) irrigation (PTRAWD), (iii) dry direct seeded rice with no standing water at all, (iv) dry direct seeding with AWD irrigation (DDSR-AWD), (v) dry direct seeded rice with 3-5 cm standing water from panicle initiation to heading, (vi) dry direct seeded rice with 3-5 cm standing water from panicle initiation to grain filling, (vii) dry direct seeded rice with 3-5 cm standing water from heading to grain filling. Two rice varieties included in the experiments are BRRI dhan28 and BRRI dhan29. The frequency and amount of irrigation for different plots were recorded. Data on different crop characters, yield and related attributes were collected and were subjected to statistical analysis using a computer package program MSTAT-C. The results reveal that rice gave the highest grain yield in dry direct seeding system where the standing water was maintained from panicle initiation stage to flowering (T5) or grain filling stage (T6) at reproductive phase. The T5 treatment required 7-9 irrigations whereas T6 required 9-12 irrigations. The treatment T5 required 50% less irrigation water than PTR-CI method. Thus, it may be concluded that dry direct seeded rice should be practiced for rice and standing water should be maintained from panicle initiation to heading for obtaining higher grain yield with least irrigation input.