Water availability for agriculture is decreasing mainly due to increasing consumption from cities and industries. But a 1% increase in water productivity in agriculture will make available an extra 24 liters a day per person. Flooded rice is taking up about 3,000 liters to produce 1 kilogram of grain and its consumption of irrigation water on earth is about 100 Mm3 year-1. This system, however, offers big opportunities to increase water productivity, defined as kilogram of grain per volume of drop water. To evaluate distinct crop establishment strategies concerned with planting methods and water saving, and genotype adaptability to these strategies, the performance of traditionally transplanted rice (E1) at 100 pl m-2 (25 kg seeds ha-1) was compared with that of wet direct-seeded (E2), dry direct-seeded (E3), and aerobic rice, all sown at 80 kg seeds ha-1, for six contrasting genotypes including hybrids and inbreds. Regardless of genotype, continuous flooding in E2 and E3 resulted in grain yield higher than that measured in E1 (22% lower) and E4 (28% lower): a lower panicle density at maturity was reported in E1 whereas a lower percentage of filled grain was reported in E4. During early crop growth, the number of tillers in E2 was significantly higher, about 650 m2 at 14 d after sowing, than that in E3, E4, and E1, about 400, 300, and 100 m2, respectively. This trend was also observed in leaf area and leaf and panicle dry weight dynamics and could be attributed to the quicker emergence of the pre-germinated seeds in E2, despite a 3-d delay in sowing. The calculation of water use in E4 was difficult in terms of estimating water outflows in surrounding canals from heavy rainfall, which were frequent during the season because of the intended absence of bunds surrounding E4. Moreover, seepage and percolation were not evaluated. In E3, water use during crop growth was high due to the heavy demand from the dry soil right after crop emergence and throughout plant growth. W